hamradiotools

Nexus 1.12.0

Released 14 September 2026. The current release is 1.24.0. Download it · Every version

Added

  • Nexus Remote stays on across a restart.

    If Remote was on when Nexus closed, it turns itself back on and reconnects when Nexus starts again — no trip to Settings, no click at the shack. If you turned it off, or revoked the station's access, it stays off. A locked credential store still leaves it off. Each browser's permissions come back with it, but only for a browser that is still approved exactly as it was: revoke a browser, or approve it again, and it starts with nothing. That includes FT8/FT4 transmit where you ticked it on the approval, so it is remembered until you revoke it. Transmit still starts off: the TX switch is off at every launch, as it always has been, and a remembered permission sends nothing until the browser presses TX On. Turn off Remote only disconnects: every browser drops at once, and turning Remote back on restores what each still-approved browser had. Revoking a browser, revoking the station's access, and "End remote control and clear permissions" still clear them.

  • Start Nexus when you sign in to this computer.

    A new switch under Settings ▸ Appearance ▸ Start at sign-in, off by default. It is what lets Remote come back after the computer itself restarts, so the first time you turn Remote on, Nexus offers to switch it on for you — once, and only if you say yes. If your computer will not let Nexus start at sign-in, Nexus says so and the switch stays off. Starting at sign-in is an ordinary start: Nexus listens, and transmit stays off.

  • Signing up for Remote is one pass, and the site looks like hamradiotools.io.

    A new account enters the six-digit code emailed during sign-up and lands signed in — no "sign-in refused" page. If a sign-in is ever held for an unconfirmed email, the site says "Confirm your email to finish" with a Continue button and a way to use a different account, and refusals that used to say "check the connection" now say what happened: that email already has a trial under another sign-in, or too many tries in a short time. The Remote sign-in, station list and pairing screens take the hamradiotools.io colours, type and Nexus mark; the Nexus screens inside Remote are unchanged.

  • Remote says "Not sent" when a command never reached the station.

    On a slow link the browser refuses to send a command against a station reading that is out of date, rather than sending it late, and says so: nothing reached the station, so nothing changed there, and to try again. "Not confirmed" means a command was sent and never answered. Stop TX is never held back this way.

  • Late station readings fade the Remote screen instead of blanking it.

    When the station's readings stop arriving for about five seconds, the browser's workspace stays on screen, faded, and can't be clicked until readings are current again; the banner says "Station data unavailable" next to the control status. Stop TX stays at full strength and clickable. Nothing can be sent from faded readings; every command still refuses data that isn't current. The banner keeps the same size throughout: after a station command the "Release station control" button and the result buttons stay in place, greyed out while the station is re-checked, the result line keeps its space, and the status says "Station control active" for as long as the station reported the control lease as valid. The frequency digits stay drawn between heartbeats and only tune once control is confirmed again.

  • Remote: dragging a level slider or typing an FT offset no longer gets cancelled when the link is slow for a moment.

    The value is sent once station control is current again, or you're told it wasn't sent.

  • Remote: when the station is busy and refuses a command, Nexus says "The station was busy and nothing changed. Try again."
  • Remote pilot: Stop TX stops anything, from every cockpit, and a slow link no longer flickers.
    • Stop TX in the browser now stops any transmission at the station, however it started — a keyed mic, the tune carrier, the CW or voice keyer, RTTY, PSK, SSTV or FT8/FT4 — and it is there in every cockpit, not only Operate. It needs station control, not transmit permission, it is never greyed out by a slow link, and it is sent at once. It leaves TX off, exactly as pressing Stop TX at the shack does, and your own Stop TX at the shack works as it always has.
    • On a slow connection the workspace holds still: controls stay usable through brief delays, and an open dropdown stays open. A change made during a delay is sent once the connection catches up, or you're told it wasn't sent. FT8/FT4 TX, CQ and Log clicks are never held back: during a delay they say "Not sent" at once.
    • "Station data unavailable" and the faded screen appear only after about 5 seconds without station data, and clear as soon as data returns.
    • Scroll-wheel, arrow-key and digit tuning no longer drop steps on a slow connection: a burst of steps is sent as one tune once the connection catches up, or refused as a whole.
  • Export one POTA activation, not a date range.

    The Logbook's export area now lists your activations — park, UTC date and contact count — and exports exactly that one. Previously the only filter was a date range, so an activation shared a day with any ordinary contacts you made and they came out in the same file. Two parks in a day is simply two rows and two files. The file is named the way POTA asks ([email protected]), and an activation that crosses UTC midnight gives you two files rather than one — POTA would accept the combined file, but credit is per Zulu day, so six contacts either side of midnight is twelve contacts and no valid activation. Two rows saying "6 QSOs · under 10" tell you that before you submit; one file does not. Hunter contacts — where the other station was in the park — can never appear in an activator file. Choosing an activation supersedes the date range rather than combining with it, so a leftover date cannot silently shorten a log you are about to upload. Not handled: a park spanning two states needs a fourth part in the filename that Nexus cannot fill, and a two-fer is not modelled — one QSO carries one park reference.

  • Sweepstakes, CQ WW and CQ WPX.

    ARRL November Sweepstakes ships as two contests (the CW and Phone weekends are separately run), and CQ WW and CQ WPX as four. These are the hard ones: unlike a QSO party, their multipliers come from the callsign rather than the exchange — DXCC entity, CQ zone, and WPX's prefix rule — so Nexus now resolves each contact's country and zone as you log it. A relation-scored contest refuses to start without a country file rather than quietly scoring a whole weekend at zero. Sweepstakes' exchange is the awkward one in amateur radio — serial, precedence, callsign, check and section, with the callsign inside the exchange and the check constant across your whole log. Both are handled, and the check is watched: change it mid-contest and Nexus tells you at log time and refuses the export naming both values.

  • ARRL VHF, all three runnings.

    January, June and September are separate entries in the contest picker because they are scored differently — January pays more for 902 MHz and up than June and September do — and each has its own weekend, ADIF contest id and Cabrillo name.

  • A rate meter on the contest scoreboard.

    Contacts per hour over your last 10 contacts, your last 100, and the last 60 minutes, for every contest in the picker, both Field Days included. Each is measured up to now rather than between logged contacts, so it drops when you stop instead of showing your best run on a dead band.

  • The end-of-contest merge has a button.

    The contest view says what it will do before you press it — "Merge 3 contacts into my logbook" — and afterwards how many were added and how many were already there, so a second press is visibly harmless. It sits beside the switch that decides whether merged contacts are queued for upload, with the one catch written underneath: ClubLog's own catch-up still sends any contact it never accepted the next time you save a ClubLog password, whatever that switch says. A club running multi-op can now declare it under Settings ▸ Contesting; single-op is the default.

  • Contest support beyond Field Day — four state QSO parties.

    Tennessee, Ohio, California and Texas are selectable in Settings ▸ Contesting, and the whole operating path follows the one you pick: the entry strip shows that contest's exchange fields, dupe checking uses that sponsor's own rule, the scoreboard counts its multipliers, and the Cabrillo and ADIF exports carry its identifiers. Every rule was read from the sponsor's current page rather than inferred, and where a sponsor's own wording is ambiguous the ruleset records the ambiguity instead of guessing at it. Two consequences worth knowing. Ohio has three roles — in-state, W/VE and DX — and which one you send follows your station data rather than being typed each time. And Tennessee and Texas both have computed bonus terms that this scoring model cannot express, so their scoreboards say plainly that the total excludes bonuses: the log is submittable, the number on screen is not your final claimed score.

  • Close and resize Connect's panes.

    Every pane on Connect has a ✕, and a Panels menu brings back anything you closed, with Undo and Reset layout. When a pane closes, its neighbour takes the space; close a whole side and the map takes the width. Drag the edge of a side rail to make it wider or narrower, or drag between its two panes (both also work from the keyboard, and a double-click puts the default back). What you close and how wide you make things is remembered per window, and a saved width is trimmed to fit a smaller screen. Nothing changes until you use it, and Reset layout returns Connect to exactly how it first opened.

  • Full-screen map.

    One button on the map toolbar hides the surrounding chrome and gives the map the window; the same button and Escape both bring it back. The choice is remembered per surface, so a popped-out map does not inherit what you hid on the main window.

  • The map draws your transmit and receive paths.

    A dashed green great circle runs from your QTH out to every station that reported hearing you, and a dotted blue one out to every station you decoded — the picture GridTracker operators know, on both the 2-D map and the 3-D globe, and correct in all three projections (it bends round the globe, stays a true beam radial on the beam map, and cuts cleanly at the antimeridian on the world map). Two things keep it a map rather than a spider's web. Each direction is its own layer with its own checkbox and opacity slider, and each is capped at 30 paths, freshest first — a contest pileup cannot bury the coastline. Paths also expire rather than piling up: a reception report older than 30 minutes and a station that has gone stale in the roster stop drawing at all. Heard-you (TX) paths are on by default; heard-by-you (RX) paths are off — the first set is a handful of paths worth seeing unasked, the second is your whole decode roster, which on a busy FT8 band is a hundred stations. One checkbox in Layers turns it on. A station heard both ways draws once, in green. The 3-D globe's existing heard-me arcs are now capped the same way; they never were.

  • Nexus Remote, an early pilot: your station in a browser.

    Pair the Nexus at your shack with an account under Settings ▸ Station ▸ Remote access, approve the browser at the radio, and the browser opens the same Nexus you use at home — the FT8/FT4, CW, Phone, RTTY, PSK, JS8, Tempo, SSTV and APRS screens, the logbook, spots, the needed list, awards and stats, DXpeditions, POTA/SOTA, memories, satellites, Field Day and the contests, and your settings. Your log, settings and radio never leave your computer: the service carries the picture and your commands and nothing else, and if it goes down nothing at the station changes. It runs on a test server that may be reset, and approving a station starts a 14-day trial, one per account. Pairing always finishes at the radio, so nobody can pair a station they are not standing at, and it takes one approval there: approving the pairing turns Remote on and approves the browser you paired from. A station can be renamed from the browser, and a browser or a whole station can be revoked from either end.

  • Operating from the browser, with permission you grant at the shack.

    Approving a browser gives it station controls and remote logging, and FT8/FT4 transmission too when you tick "Also allow FT8/FT4 transmit" on that approval; the switches under each browser then take one permission away or give it back. Station controls cover tuning (a typed frequency, the dial digits, the scope wheel, a click on a signal), band picks and "Use this mode", the Phone screen's AUTO/USB/LSB/FM/AM with repeater shift and tone, NB, NR, notch, AGC, filter width, RF power and mic gain, decode depth, RX offset, JS8 speeds and the MSK144 period, arming the RTTY, PSK, SSTV and APRS decoders, the amplifier's Operate/Standby, band steps and band-following, and which radio is active. Remote logging lets the browser's Log QSO form write to your logbook; Field Day contacts still log at the shack. One browser holds control at a time, a control only shows as done once the radio has confirmed it, and a command whose outcome is uncertain is never repeated on its own.

  • FT8 and FT4 transmit from the browser.

    With station control and the transmission permission, the browser can call CQ, answer from a decode or the roster, step through Tx1–Tx6, resend, send free text and set the TX offset and period. It goes through your normal TX switch and TX watchdog, Stop TX works from the browser, and if the browser goes away the station stops transmitting within five seconds. Nothing else transmits remotely — there is no remote PTT, CW keying or keyboard-mode sending. Not yet: no audio reaches the browser in either direction, and most of the radio controls have not yet been checked against a real radio.

Fixed

  • HRDLog.net uploads work.

    Every upload to HRDLog.net has failed, on every computer, since the connector shipped (#224). HRDLog's server only offers encryption settings that Nexus's built-in secure-connection library does not support, so it hung up before anything was sent. Nexus now uses your operating system's own secure connection for HRDLog.net, and only for HRDLog.net; every other service is unchanged. The certificate is still checked, TLS 1.2 or newer is required, and your upload code stays in the system keychain and never appears in an error or the connection log. If that connection is dropped while it is being set up, Nexus now says so, instead of blaming your antivirus or a proxy. A failed HRDLog upload is retried twice, as HRDLog asks of logging programs, instead of up to twenty times. Uploads also tell HRDLog which version of Nexus sent them. On Linux, Nexus now needs OpenSSL 3 (libssl3). The .deb packages install it for you, and the AppImage carries its own copy.

  • Connect's map remembers each intent's setup.

    Moving from Chase DX to POTA/SOTA and back used to reset Chase DX to its preset, because every intent shared one projection, one set of layers and one colour mode. Each intent now keeps its own, including 2-D or 3-D; a preset applies only the first time you use an intent, and whatever you had set before this change is kept for the intent you were on.

  • One map picker: Globe, 3D, Flat or Beam.

    Connect had a 2D/3D switch and, inside 2D, its own Globe/Beam/World buttons — so leaving the 3-D globe could land on a 2-D globe that looked the same, and the switch seemed to do nothing. The four views now sit in one row in the same place whichever is showing; Globe is where a new view starts, and each view remembers the one you picked. Layers sits top-left on every map and folds away the same way Conditions does.

  • The map costs far less CPU.

    The 3-D globe drew every frame even with nothing moving; it now draws only when something changes, and wakes on drag, zoom, spin, new data or a resize (the arc dashes and your station's ring pause while it is idle). The 2-D map redrew itself, country outlines and all, every time new data arrived, even when nothing on it had changed; it now redraws only when what it shows changes. With every layer off, both now sit near zero while idle.

  • Every map toolbar control is reachable at 1024×768.

    The toolbar could not wrap, so on a small window the map drew wider than its space and Reset, Full screen and the LIVE badge were cut off, and the Layers panel could sit entirely off screen. The toolbar now wraps onto a second row.

  • A long over went silent partway through and left the rig keyed.

    WSPR transmitted for about twenty seconds, the audio stopped, the power meter fell to zero — and the radio stayed on the air for the remaining ninety seconds of the over, every over. Reported on an FTdx10 under Ubuntu. Nexus hands the sound card a whole transmission at once, and the card's buffer held twenty seconds of it. An FT8 over is thirteen seconds and fitted; a WSPR over is a hundred and twelve and did not, so four fifths of it was thrown away before a single sample played, while the transmitter was held up for the length of the over Nexus meant to send. The transmission is now fed to the card as it plays instead of all at once, so the whole of it goes out and the transmitter drops when the audio really ends. WSPR was the report, but it was never only WSPR: JT65, Q65 at 30 s and longer, FST4 and FST4W at 30 s and longer — FST4W-1800 held the transmitter up for twenty-nine minutes per over. On the digital modes SSB suppresses the carrier, so once the audio stopped there was no RF at all — which is why the reporter's power meter read zero while the radio was still keyed. On FM, AM and CW it would have been an unmodulated carrier. It also caught PSK31: a one-shot send longer than about eighty characters — two lines — went out truncated with the transmitter held up for the whole message. Continuous (latched) PSK31 was never affected, because it feeds the card as it types. FT8, FT4, FT2, MSK144 and the 15-second Q65 and FST4 periods always fitted and were never affected. SSTV, CW, RTTY and the tune carrier feed the card as they go and were never affected either.

  • Nexus no longer switches your radio to CW-R on 40 m.

    Moving to 40, 80 or 160 m put the rig in reverse CW, and moving back to 20 m put it in normal CW — on every band change, undoing it if you set it back by hand. Nexus now asks for plain CW on every band, so your radio stays in whichever CW its own menu defines, and a new Reverse CW (CW-R) checkbox under Settings → CW turns the reverse mode on everywhere if you prefer it. It is off by default. This was deliberate and it was wrong. Nexus was applying the 40/80/160 m LSB convention to CW, and that convention is real for voice but does not exist for CW — CW and CW-R transmit identically, so the station you work cannot tell the difference; the only thing that changes is the pitch you hear when you tune past a signal. Worse, the two words do not mean the same thing on every radio. On a Yaesu, reverse CW is CW-L, the lower side. On an Icom it depends on that rig's own CW Normal Side menu item, which ships set to LSB — so an IC-7300 asked for reverse CW goes to the upper side, the exact opposite of what the rule intended, and its normal CW on 20 m went to the lower side. Thanks to the operator who reported it from an IC-7300 on 40 m under Linux. The top bar now flags CW against CW-R as a mode mismatch instead of treating them as the same mode. That collapse is why none of this was visible inside Nexus and had to be spotted on the radio's own display. If you turn reverse CW on and your rig has no reverse CW at all (FlexRadio SmartSDR, PowerSDR, Thetis), expect a standing mismatch pill: Nexus falls back to plain CW on those and the pill is telling you so. The soundcard keyer is unchanged. It keys an audio tone through a data mode rather than putting the rig in CW, so the sideband there really does follow the band, exactly as it does for FT8.

  • An AM contact was logged as SSB.

    Work someone on AM from the Phone screen and Nexus wrote the QSO to your logbook as SSB — including on 14.286, the 20 m AM calling frequency. The Phone screen had only two answers for what mode you were on, FM and SSB, and everything that was not FM came out SSB. It now logs AM as AM, and the entry strip's "Logs to the shared logbook as …" line says so before you commit the contact. The log follows your radio, not what Nexus asked it for. If the rig is on AM, the contact is AM, whatever the Phone screen last commanded — which is how this started: the Phone screen had no AM button on 20 m, so the reporter set AM with the radio's own mode knob. It cuts the other way too: pick AM and your rig does not take it, and the contact logs as the SSB it really was. Nexus believes a mode read-back only while CAT is up and the radio has actually answered, and a radio sitting in CW or a data mode names nothing this screen can log from, so that keeps the commanded mode as before. Everything downstream carries AM once the log does — ADIF export, the N1MM and N3FJP feeds, and your uploads to QRZ, LoTW, ClubLog and eQSL. This fixes contacts from here on; an AM QSO already logged as SSB is yours to correct in the Logbook. AM is on the Phone screen's mode picker on every band now. It was hidden below 10 MHz and from 28 MHz up, which left no AM button on 14.286 — the one frequency it most needed. USB, LSB and FM never had a band filter, and a radio that cannot do AM somewhere still refuses it itself.

  • Your ARRL section list was eight years out of date.

    Nexus carried 83 sections including MAR, GTA and NT. ARRL now publishes 85: the Maritime section split into NB, NS and PE, GTA became GH and NT became TER. So Field Day offered three sections ARRL does not recognise and was missing five it does — and a log submitted under a retired code is not a log ARRL can score. GTA and NT are renamed for you, because ARRL publishes those transitions itself. MAR is not: it split three ways and only you know which one you operate from, so Nexus keeps what you had, says what changed, and asks. It does not guess, and it does not blank the setting — that would destroy the only record of where you were.

  • Your transmit timing was being steered by a single unchecked network packet.

    Nexus measures the PC clock against a time server every ten minutes and shifts its own transmit and decode windows by the result, so slots land on the true UTC grid. That measurement used to be one reply from whichever server answered first, held for ten minutes with nothing to corroborate it — a captive portal, a server mid-restart or a hijacked reply would have moved every over you sent. It now asks three servers, requires at least two to agree within 200 ms, uses the median of those that do, and confirms the whole thing with a second round a few seconds later. If nothing agrees, nothing is published.

  • The clock correction no longer disappears the moment you lose signal.

    Walk out of coverage and the timing correction that had been keeping you decoding used to vanish silently on the first missed measurement — worst on exactly the portable stations that needed it. The last good measurement is now held with an age, for a window derived from how fast your machine's clock has actually been observed to drift, and the top bar tells you when it has aged out instead of the correction simply stopping.

  • A clock more than a minute out is no longer corrected silently.

    Nexus refuses to steer by an offset that large and says so instead. Slot timing is not the only thing a clock that wrong breaks: your logged QSO times come from the machine's clock, so they would be wrong by the same amount in every record you upload.

  • Waking a laptop from sleep no longer leaves the wrong correction applied for up to ten minutes.

    Windows restores the system clock from the hardware clock on resume — a jump of hours, on a real machine — and the correction measured before that describes a clock that no longer exists, so applying it makes the timing wrong the other way. Nexus now notices the jump within a tick, drops the stale correction, re-measures immediately, and on Windows asks the time service to re-check rather than waiting for its next scheduled poll. If the jump lands mid-transmission the over is ended rather than held: the transmit watchdog measures elapsed wall-clock time and cannot see a clock that moved backwards, so a keyed radio would otherwise have stayed keyed.

  • Amplifier band-following shows and uses the choice you saved.

    The follow-band checkbox in Settings and the radio profile underneath it could disagree, so after a save, a restart or a switch of radios the box could say one thing while the amplifier did another. They are now kept in step both ways, and choices you had already saved are preserved.

  • Amplifier band-following checks before every command it sends.

    Nexus re-reads your setup before each serial write to the amplifier, so turning follow off or changing the port mid-poll cancels the pending step, a button you press wins over an automatic band step, and nothing is sent while you have the radio keyed by hand. The KPA waits for a fresh reading that the radio is idle, and SPE amplifiers stay within the model's band limit. Not yet verified with a real amplifier.

  • A radio that will not connect keeps backing off.

    A background radio whose connection failed now keeps waiting longer between retries, up to a minute, instead of starting over; correcting its connection settings retries straight away; and switching to a radio that is still connecting waits for that attempt rather than opening a second one against the same port.

  • The RTTY sequencer's sign-off sent the next contact's serial number.

    The counter moved on when a contact was logged, and the sign-off went out after that. A serial is now issued once, when the exchange is first sent to a station, and every repeat to that station — the re-call, an AGN — carries the same number. An abandoned contact's number is skipped rather than reused, because a gap is invisible to log checkers and a number sent twice is an error against both stations.

  • A Windows sound-card glitch no longer floods the log, and receive audio gets more buffer.

    One operator's five-hour log was 61% capture stream died — 535 lines — while nothing had died: that message is Windows reporting a single late packet and delivering it anyway. Nexus had been asking Windows for its smallest capture buffer, one 10 ms device period, and now asks for about 100 ms of headroom (it adds no delay). The message is rate-limited with a count and no longer calls a glitch a death, and the "no decodes" line now says whether audio is reaching the decoder at all, which separates a dead band from a dead audio path. Not yet measured on a Windows machine.

  • Pop-out panels scroll again.

    A torn-off POTA/SOTA board could not be scrolled — content below the fold was simply unreachable. The pop-out window had no scrolling container at all, an omission since pop-outs were introduced; every other pop-out was checked and only this one was affected.

  • A solo Field Day entry no longer claims to be multi-operator.

    The Cabrillo header hardcoded CATEGORY-OPERATOR: MULTI-OP, so every single-operator log Nexus has ever exported submitted a false claim about the station. It now follows your entry category, and a club host's merged export still says MULTI-OP, because that one genuinely is.

  • Three Cabrillo contest names were wrong.

    Field Day, Ohio and Texas were submitting headers under names their sponsors do not use — ARRL-FD, MRRC-OHQP and TXQP are the correct ones. The ADIF CONTEST_ID tag keeps its own separate values, which were right all along; the two are different registries that happen to agree for most contests and disagree for these three.

  • The beta-updates switch could silently turn itself off.

    Turning it on and then touching any APRS control wrote a stale copy of your settings back over it, so you would quietly stop being offered betas with nothing to indicate why. The stale write is fixed at its source, which protects every other setting that surface could have reverted.

  • Interop broadcasts carried the wrong exchange for earlier contacts.

    The WSJT-X and N1MM feeds stamped your current exchange on every contact in the batch, so a mobile station that changed county mislabelled everything it had already worked. Each contact now carries the exchange it was actually sent with. Field Day operators see no change — their exchange never moves.

  • N1MM broadcasts named the wrong contest.

    Anything that was not Winter Field Day was announced as ARRL Field Day, which was harmless while those were the only two contests and wrong the moment a QSO party could be selected.

  • Map icons are sharp again at UI scales above 100%.

    The map canvas was sized in layout pixels while being painted in device pixels, so any zoom above 100% upscaled a too-small bitmap. Station and spot markers also scale with the map surface now, with a dark outline behind each dot and label so they stay legible over coastlines, and the hover card is anchored to the marker instead of chasing the cursor.

  • Correcting a busted callsign now re-sends the QSO to the logging services.

    Fixing a call in the Logbook's edit form has always cleared that contact's upload stamps — the point being that QRZ, ClubLog, eQSL and the rest hold the old call and need to hear the correction. But nothing ever put the corrected contact back on the upload queue, so they never did: your log showed the right call and every service kept the wrong one, permanently, with the stamps that were the only sign anything was owed now erased. A callsign correction goes back out to every connector you have switched on. Other edits (name, grid, RST, a park reference) still don't re-upload — the services already have that contact under that call.

  • The Logbook edit form's CALL box shows the whole callsign.

    It shares its row with the QRZ lookup button, and in a normal-width window that left it about five characters wide, so a six-character call read back as if its last letter had been dropped. The value was never short — what you saved was always what you typed — but you could not see it. The button now steps down to its own line when the field gets tight.

Changed

  • The clock chip says what Nexus is doing, instead of telling you to go and fix it.

    Nexus has been correcting the PC-clock offset itself since June; the top bar had not caught up, and painted a red alarm and a "sync via NTP / time.is" tooltip on a station whose timing was in fact correct. It now distinguishes four things: the offset it is already applying (with how old that measurement is and how many servers agreed), a correction that has aged out because nothing has answered, a clock too far out to correct, and — the one case where it asks you to do something — the machine whose clock you need to set yourself. It also names who owns the clock: if you run NetTime, Meinberg or Dimension 4, Nexus deliberately touches nothing and now says so, rather than leaving you to wonder why.

  • Nexus diagnoses its own clock problems, and on Windows fixes the ones it can.

    A time service that a "debloat" script disabled is started; one that is running but has never reached a server is told to re-discover and re-sync; a healthy machine that only checks the time every nine hours is asked to check every seventeen minutes. Each needs one administrator prompt, once, and only on a machine a check has already found to be broken — a healthy machine is left completely alone, and so is any machine running a third-party time client. Nexus never changes which time server your machine uses. On Linux and macOS nothing is changed at all: their time daemons already check far more often than Windows does.

  • Nexus's own time check no longer uses the NTP Pool.

    The pool is volunteer-run and its vendor policy asks applications not to use its default zones as a shipped default; Nexus now asks operator-run public time services instead, and spreads its check times so a thousand installations do not arrive on the same second.

  • Xiegu radios: trust the radio's own SWR meter.

    Nexus reads CI-V transmit meters on an Icom scale, and a Xiegu's meter uses its own — a G90 showing 1.2:1 on its front panel can read 6:1 in Nexus. The Xiegu rig guide now says so plainly; a real calibration needs bench measurements first.

  • Field Day refuses a section its own rules do not list.

    Entering an unrecognised section used to start the mode anyway and let you transmit it; it now declines, the way a blank section already did. If you operate a section Nexus does not know, this will stop you where it previously did not.

Every version