Upgrading to NereusSDR 2026.10.1

Use this note when updating a station from 2026.10.0. It lists what changes when you first launch, what to do on a Core, and a short list of things to try. If you are coming from 0.5.2, read the 2026.10.0 checklist first. When reporting a problem, include the installed version, operating system, radio model and firmware, and whether you use a local or remote Core.

Update the Core and the desktop together

The Core owns the radio, the station DSP and the accessories. The console runs where you operate. Update both. Rotor control, the CAT Control page on the phone and the Core speaker need a 2026.10.1 Core. An older Core grays them out with a reason such as "This Core does not control a rotor. Updating the Core may help.", and older Cores, desktops and phone apps still connect to each other. The phone shows the new Display pages only with a 2026.10.1 Core; with an older Core it keeps the earlier pages.

What changes on first launch

Keep a backup of your settings before the first launch, and close the console normally so the new version can migrate them once.

• Two-tone is one setting. The two-tone level (Test > Two-Tone IMD) is now a single value for the radio, and TX profiles no longer save any two-tone values. If your saved level was the old default of -6 dB, it becomes 0 dB. That is 6 dB more two-tone drive than before for anyone who never changed it, and it lets PureSignal calibrate. If you use PureSignal, check your two-tone level and drive before the first calibration. Any saved level other than -6 dB is left alone.
• Audio choices move to the new Driver list. Each device card now has one Driver list above Device, replacing the Driver API list and the Windows audio checkboxes. Your saved choices carry over once, and the old settings are kept. On a headless Linux Core, every profile moves to ALSA direct. After the first launch, open File > Settings... > Audio and confirm that your speaker and microphone picks are still the ones you expect.
• The workspace look is the default. Containers dock on all four edges, headers show or hide with an eye button, and there is no container Lock with this look. Locks you saved earlier open unlocked. Your saved layout is copied the first time you start, so nothing is lost.
• On the Mac, the Mic Mode goes to Standard. NereusSDR sets its own Mic Mode to Standard at each launch, which ends the long wait for the microphone after launch. If you picked Voice Isolation or Wide Spectrum for NereusSDR in Control Center, it is back at Standard. The first window can appear up to about a second later.
• Attenuator and preamp are remembered. They are saved and sent to the radio each time you connect. Connecting never keys the radio.
• Each receiver keeps more of its own settings. L/R pan, binaural, squelch, APF, the FM settings, the DIGL and DIGU offsets and the RTTY mark and shift are now saved for each receiver, so they start saving from the first launch. AGC-T AUTO and its offset move from each band to each receiver: the first time, a receiver takes the value it had saved for the band it opens on. Receivers B, C and D now run on the settings they show, so one whose saved settings differ from the defaults may sound different from before.
• FM squelch reads 0 to 100. Squelch Threshold on File > Settings... > DSP > FM now shows 0 to 100, where 100 is the tightest, instead of dB. 2026.10.0 did not save this threshold, so every receiver now starts at 100, the tightest. With FM squelch on it now opens; lower the threshold if it stays closed.
• Display settings are now defaults plus per-pan values. Setup's Display pages (now Spectrum, Levels, Waterfall, View, Peaks & Markers, FFT & Engine, Transmit Display and Multimeter) set the default for every pan, and each pan's Display pill can keep its own value, marked with an amber dot. Old page names such as Spectrum Defaults and Grid & Scales no longer exist. Reset this pan returns one pan to the defaults.
• Waterfall Black Level and Color Gain are converted once. With Clarity, Black Level now runs 0 to 100 with 50 at the noise floor, and Color Gain sets how many dB the colors cover; in the other choices they trim Low and High. Your saved values, and each pan's, are converted so the waterfall looks the same as before. If it looks different, check Levels.
• Cal Offset is reset and stored noise floors are cleared once. The saved display Cal Offset values (the Setup default and each pan's) held the station's offset, so they are removed and Cal Offset starts at 0 dB; the control stays on Setup's Spectrum page. The stored per-band noise floors that Clarity measured before calibration are removed, so Clarity measures each band again the first time you tune to it. Your radio's own calibration is not touched.
• FFT size has a Sizing choice. Sizing is Auto, Hz/bin or Fixed. If you had a Hz/bin target above zero, you start in Hz/bin. Otherwise you start in Auto, and the FFT Size you saved applies only if you choose Fixed. Check FFT & Engine if your spectrum looks coarser or finer than before. The setting is shared by every pan and every window connected to the Core.
• AGC Off is quieter. With AGC set to Off, a receiver now runs the Fixed Gain on File > Settings... > DSP > AGC/ALC, 20 dB unless you changed it. In 2026.10.0 it ran 60 dB until you changed the value, so AGC Off is about 40 dB quieter. Raise Fixed Gain if you want the old level.
• Every receiver starts with your DSP Options sizes. Receivers B, C and D did not take the block size, filter size and filter type on File > Settings... > DSP > Options until their next mode change, and ran a 4096-sample block with a Linear Phase filter until then. They now use the saved values for their mode, 64, 4096 and Low Latency unless you changed them, so they may sound and respond differently. The FM row on that page now starts at 64 for receive and transmit.
• PureSignal looks different. The applet and the Bench show CORRECTION and CALIBRATION FEEDBACK with a class (Too low to High), Advanced now hides the bottom strip too, and AmpView names its curves differently. Two-tone IMD readouts on the display are off until you check Show 2Tone measurements.

Core installs on a Raspberry Pi, Rock or other SBC

The Core runs as a service account with no groups of its own, so Linux sound cards and serial adapters each need a small systemd drop-in.

• Audio. Every SBC install and every .deb package now ships or writes the audio drop-in. An older install that lacks it cannot open a sound card, so the Core speaker will not work until you add /etc/systemd/system/nereusd.service.d/audio.conf with [Service] and SupplementaryGroups=audio.
• Serial. A rotor on a GS-232 adapter, or serial CAT on the Core, needs /etc/systemd/system/nereusd.service.d/serial.conf with [Service] and SupplementaryGroups=dialout. The Pi 4 and Rock installers and the station image write it; the .deb packages do not. After adding either file, run sudo systemctl daemon-reload and sudo systemctl restart nereusd.
• Rock real-time priority. On a Rock, run the 2026.10.1 installer once and then restart the board. It adds a drop-in meant to keep an ssh login from costing the Core its real-time priority. See What to try.
• Network receive buffer. The 2026.10.1 packages and the Pi 4 and Rock scripts raise Linux's net.core.rmem_max to 4 MB through /usr/lib/sysctl.d/10-nereusd.conf, and never lower a limit that is already higher. Removing the package removes it. The README section "The system setting the Core's installer changes" says how to check or undo it.

The SBC install guide has the details, including the sound card's own mixer and kernels that refuse real-time priority. Building the PulseAudio engine from source on Linux needs libpulse-dev.

What to try

Each item is a short check you can do in a few minutes. Note the result, including anything that surprised you.

1. Rotor. Open File > Settings... > CAT & Network > Rotor, choose your controller (Yaesu GS-232A, serial port, Yaesu GS-232B, serial port, Hamlib rotctld, already running or Hamlib rotctld, started by the Core), set the end stop and travel, and select Save and connect. Open the Rotor applet, turn to a heading, stop it, and try a preset. Right-click a spot and choose Turn beam to the call. On the iPhone or iPad, open the Rotor page and confirm it shows the same heading. Release a held turn button and confirm the rotor stops.
2. CAT. Open Tools > CAT Control... (or the CAT pages in File > Settings... > CAT & Network), assign a channel to a slice, enable one listener, and connect a logging or digital-mode program. Change frequency from the program and from NereusSDR and confirm both follow. Try the Hamlib rigctld port with a program that speaks it. On a headless Core, set CAT up from a connected desktop or from the phone's Tools tab. Confirm that a refused setting says why.
3. Native audio. Check that the Driver lists show your system's audio system and that devices appear and disappear as you plug them in. With a Core, pick the Core speaker in Audio > Outputs and confirm the level and mute match the phone's Sound panel. Disconnect and reconnect your microphone and confirm transmit audio returns.
4. Workspace look. Hide and show a container header with the eye button. Hold Shift (or press Ctrl+Shift+A) to arrange, dock a container on each edge, pop one out, pin it and return it. Save a workspace profile and switch to it with Containers > Manage Workspace Profiles…. Choose Reset Default Layout and confirm the S-Meter comes back.
5. Receivers after a relaunch. Give two receivers different squelch, L/R pan, APF and AGC-T AUTO settings, close NereusSDR normally, open it again and confirm each receiver comes back as you left it and sounds the way it did.
6. Hermes Lite 2 through a Pi Core. Key the radio from a window or the phone through a Core on a Raspberry Pi and hold it for half a minute. Confirm the radio stays keyed for the whole time.
7. Rock real-time priority. After installing on a Rock and restarting it, log in over ssh and run cat /sys/fs/cgroup/cgroup.subtree_control. Confirm that cpu is not in the list, and that the Core's log has no "Raised thread priority was refused" line from after the restart.
8. Two-tone and PureSignal. If you use PureSignal, run a two-tone test and a calibration at your new two-tone level and compare it with what you expect. Then check Show 2Tone measurements (Bench > IMD measurements), turn on DUP, run a brief two-tone test and read WORST IMD3 and WORST IMD5 on the rail's Inspect readout.
9. RADE waveforms. In Containers > Edit Container, add Microphone waveform and Decoded speech waveform to a container whose Slice is the transmit slice and is in RADE-U or RADE-L. Speak while transmitting in RADE and confirm the microphone trace moves; receive a RADE signal and confirm the decoded speech trace moves. On a remote desktop, confirm an older Core shows a reason instead of a trace.

Reporting

Report expected behavior, actual behavior, a minimal reproduction, desktop and Core versions, platform, radio model and firmware, the connection path and a support bundle. Keep credentials and pairing tokens out of public reports. Bench results for physical serial ports, a physical rotor and logging programs on Windows and Linux are still open. Say which hardware and program you used.