Nostr notes by HamRadio.my (Unofficial)

Pi Streamer: Live Stream Your sBitx to YouTube from a Raspberry ...

2026-11-05T11:18:52Z

Pi Streamer: Live Stream Your sBitx to YouTube from a Raspberry Pi

Prabakaran VU3DXR has released Pi Streamer v1.8.1, a lightweight utility that turns the Raspberry Pi inside your sBitx into a YouTube Live broadcast station. It captures the radio touchscreen display and receiver audio, combines them with FFmpeg, and sends the result to YouTube over RTMP or RTMPS, all controlled from a simple web dashboard with no second computer and no manual FFmpeg commands.
TL;DR: Pi Streamer is a free Raspberry Pi utility by VU3DXR that live streams an sBitx screen and receiver audio to YouTube via FFmpeg and a web dashboard. Download the v1.8.1 ZIP, enter your YouTube stream key, press Start, and your HF session, waterfall and digital modes go out live.
Pi Streamer by VU3DXR: live stream an sBitx screen and receiver audio to YouTube Live from a Raspberry Pi. Source: vu3dxr.in.
What Pi Streamer Is
Pi Streamer is a lightweight live streaming utility for Raspberry Pi that captures the sBitx touchscreen display and receiver audio, muxes them with FFmpeg, and broadcasts the result to YouTube Live through a web based control dashboard.
The project was built with the sBitx transceiver environment in mind. The sBitx is itself a Raspberry Pi 4 running Linux bolted to an HF RF deck, with a 7 inch touchscreen, built in FT8 engine, logger, telnet cluster client and web interface, sold by HF Signals as a $399 assembled rig or $229 basic kit. Since the computer is already inside the radio, VU3DXR reasoned that the radio should be able to stream itself, and Pi Streamer is the result. The package is distributed as a ZIP with installation instructions, currently at version 1.8.1, and it is shared freely with the amateur radio community.
The appeal is practical rather than glamorous. Showing another operator what your sBitx is doing means showing the screen: the frequency, the waterfall, the FT8 decodes scrolling past, the filter shape you just narrowed. A phone pointed at the display never does that justice, and a full OBS Studio setup on a second computer is overkill for a club demo or a casual HF session. Pi Streamer sits between those extremes, using the Pi that is already there.
Why Streaming an sBitx Matters
Most HF operating is invisible to anyone not sitting in the chair. A contact logged as “worked JA on 15m FT8” tells a newcomer nothing about how the band looked, how the decode window filled, or what the operator actually clicked. Live video of the radio screen closes that gap, and for the sBitx the gap is wider than usual because so much of the radio lives in software.
Concrete situations where a live sBitx screen helps:
Club nights and license classes. Members who can’t attend in person watch the waterfall, the filter adjustments and the logging flow in real time instead of reading about them later.
Digital mode demonstrations. FT8 and other modes are screen native activities. Watching decodes appear, seeing a CQ answered and the exchange logged, teaches faster than any slide deck.
Remote mentoring. An Elmer can watch a newcomer tune, spot the mistake in filter width or drive level, and talk them through it over a second channel.
Special event and field stations. POTA, SOTA and contest operations gain an audience beyond the logbook, and the recording doubles as documentation of the activation.
Homebrew show and tell. Modified sBitx builds, custom skins and volunteer firmware improvements from the community fork are easier to evaluate when viewers can see them running.
None of this replaces operating skill, but it lowers the cost of sharing what the radio is actually doing.
How Pi Streamer Works
The streaming arrangement is a short pipeline with four stages. Understanding each stage helps when something breaks, because each stage fails in its own characteristic way.
Stage
What happens
Typical failure symptom
1. Screen capture
The sBitx touchscreen framebuffer or X11 desktop is grabbed as a live video feed
Black frame or frozen image, usually a display server or permissions issue
2. Audio capture
Receiver audio is taken from the available Pi audio source
Silent stream with good video, usually a wrong input device or muted source
3. FFmpeg encode and mux
FFmpeg encodes video and audio and combines them into one stream
High CPU load, dropped frames, usually resolution or bitrate set too high
4. RTMP/RTMPS delivery
The muxed stream is pushed to YouTube Live using the stream URL plus stream key
Stream never appears on YouTube, usually a wrong key, expired key or blocked port
Viewers on YouTube see the radio display and hear the receiver audio together. So an FT8 session arrives as the full picture: waterfall on screen, tones in the audio, decodes scrolling in the window.
A few technical notes worth knowing. RTMP on port 1935 is the classic YouTube ingest path, while RTMPS wraps the same protocol in TLS on port 443 and passes through restrictive firewalls more reliably. FFmpeg does the heavy lifting in both cases, which is why the same tool that records contest audio can also feed a live broadcast. The sBitx runs its SDR processing on the same Pi 4 that now also encodes video, so streaming resolution and frame rate aren’t free. Budget headroom accordingly, more on that below.
Key Features at a Glance
Live sBitx screen streaming. Broadcasts the actual touchscreen display to YouTube Live so viewers see the frequency, waterfall, operating activity and mode windows exactly as the operator sees them.
Receiver audio streaming. Carries the received signals alongside the picture, which is what makes a demo intelligible. A waterfall without audio is a silent film.
Web based dashboard. Groups streaming destination, stream key, video and audio settings and session controls in one graphical page. Routine streaming needs no terminal FFmpeg invocations.
YouTube RTMP/RTMPS configuration. The streaming URL and stream key are entered through the dashboard, making it easy to change destinations or rotate keys.
Start and Stop controls. A broadcast begins when operating begins and ends when it ends. No need to stream around the clock.
Adjustable video and audio settings. Resolution, frame rate, audio bitrate and related parameters adapt the stream to available uplink bandwidth and Pi capability.
Streaming activity monitoring. The dashboard reports on the session while it runs, so a failed push or stalled encoder is visible immediately rather than discovered from viewer comments.
What You Need Before You Press Start
The hardware list is short, but each item has a catch worth checking in advance.
An sBitx (or another Raspberry Pi desktop). Development and testing focused on the sBitx environment. Other Pi desktop applications may work, subject to their screen capture and audio routing. See the compatibility notes below.
Reliable uplink bandwidth. YouTube recommends roughly 3 to 6 Mbps sustained uplink for a stable 720p software encoded stream, and less for 480p. Test with a speed test from the operating location, not from memory. Hotel, field and shared club WiFi often delivers far less upstream than the advertised headline rate.
A YouTube account cleared for live streaming. A new live account typically waits about 24 hours after first enabling the feature before YouTube activates it, and streaming from a verified account avoids mid session interruptions. Create and verify the account well before the demo night, not during it.
Your stream key, kept private. YouTube Studio generates the key under Create then Go live. Anyone holding the key can broadcast to your channel, so treat it like a password, rotate it after public demos, and never paste it into screenshots or shared config files.
Wired network if possible. The sBitx supports WiFi and its web interface works over it, but a wired Ethernet connection to the same router removes the most common source of dropped frames in a live RF environment full of switching supplies.
Setting Up: From Download to First Broadcast
Pi Streamer ships as a ZIP containing the application and installation instructions. The outline below follows the documented flow. Exact file names and screens come from the package README, so open that file first and let it overrule this summary if the two differ.
Download the package. Grab Pi_Streamer_v1.8.1.zip from the project page onto the Pi or a machine that can copy files to it.
Read the included instructions. The ZIP contains the install steps for the sBitx environment. Follow them in order, including any dependency installs, before launching anything.
Enable live streaming on YouTube. In YouTube Studio, open Create then Go live, complete the verification if this is a first stream, and note the 24 hour waiting period for brand new live accounts.
Copy the ingest URL and stream key. YouTube shows an RTMP or RTMPS server URL plus a unique key. Keep this window private and never screen share it during the broadcast.
Configure the Pi Streamer dashboard. Enter the streaming URL and key, pick a conservative starting point such as 720p at 30 fps with a 128 kbps audio track, and save.
Do a private rehearsal. Set the YouTube broadcast to Private or Unlisted, press Start in the dashboard, and watch from a phone on mobile data, not the shack WiFi. Check lip sync between waterfall events and audio, confirm the touchscreen text is readable at 720p, and verify CPU load leaves the radio responsive.
Go public. Switch the YouTube broadcast to Public or schedule it, announce the session to the club, and keep the dashboard monitoring visible on a corner of the screen.
Watch the official demonstration to see the finished result before installing:
Prefer the YouTube page directly? Watch the Pi Streamer demonstration on YouTube.
Tuning Quality: Resolution, Frame Rate and Bitrate
The Pi 4 inside the sBitx is simultaneously running the SDR, the touchscreen UI and now a live video encoder. That shared load is the central tuning problem. Start conservative, verify stability over a full operating session, then raise quality one step at a time.
Target
When to use it
Suggested starting point
480p at 24 to 30 fps
Field operation, weak uplink, congested club WiFi
Video around 1 to 1.5 Mbps, audio 96 to 128 kbps
720p at 24 to 30 fps
Home station with 5 Mbps or better sustained uplink
Video around 3 to 4.5 Mbps, audio 128 kbps
1080p
Only with strong uplink and a cool Pi, and only if viewers actually need full HD text
Video 6 Mbps and up, watch CPU temperature closely
Practical rules that save debugging time:
Lower the frame rate before lowering resolution. Radio screens change slowly. Dropping from 30 to 24 or even 15 fps cuts encoder load sharply while waterfall text stays readable.
Prefer wired Ethernet. A cable eliminates WiFi contention with the microwave oven, the club projector dongle and twenty phones in the room.
Mind Pi thermals. Sustained encoding on a Pi 4 in a warm shack throttles the CPU. A small fan or heatsink case pays for itself the first time a two hour net doesn’t degrade into slideshows.
Set audio for speech and tones, not music. 96 to 128 kbps AAC carries SSB audio, CW sidetones and FT8 warbles cleanly. Higher audio bitrates steal bandwidth from video with no audible benefit on a receiver feed.
Test at operating temperature. A five minute test on a cold Pi proves nothing. Run the full rehearsal length and watch the dashboard for climbing CPU or dropped frames in the final quarter.
Beyond sBitx: GQRX, SDR++, OpenWebRX, WSJT-X, JTDX and GNU Radio
Although Pi Streamer was built and tested against the sBitx, the underlying approach of screen capture plus FFmpeg plus RTMP applies to any Raspberry Pi desktop application. VU3DXR names several plausible candidates:
GQRX: live demonstrations of SDR reception and spectrum activity, ideal for showing filter, gain and antenna comparisons side by side with narration.
SDR++: sharing software defined radio experiments and receiver operation, including multi VFO layouts that photograph poorly but stream well.
OpenWebRX: broadcasting receiver activity and spectrum displays from a monitoring station, useful for club spectrum nights.
WSJT-X: demonstrating FT8 and other digital modes with the full decode window visible to students.
JTDX: sharing digital mode reception and decoding activity as an alternative WSJT-X style client, popular with sBitx operators who run it directly on the radio.
GNU Radio: demonstrating SDR experiments and signal processing projects, where the flowgraph and the resulting waterfall tell the story together.
Compatibility note, stated plainly: these alternative applications haven’t all been individually tested with Pi Streamer. Screen capture paths differ between X11 and Wayland sessions, audio routing depends on whether the app uses PulseAudio, PipeWire or direct ALSA, and per application setup will vary. Budget an evening of testing per application, confirm audio routing with a private stream first, and report results back so the compatibility picture improves.
Pi Streamer vs Other Ways to Get Your Shack on YouTube
Streaming a radio screen isn’t new. What Pi Streamer changes is where the work happens and how much gear it takes.
Approach
What it needs
Strengths
Weaknesses
Pi Streamer on the sBitx
sBitx or Pi, network, YouTube key
No second computer, dashboard instead of FFmpeg CLI, screen plus receiver audio together
Tied to Pi encoder capacity, alternatives to sBitx need manual audio setup
OBS Studio on a second PC
Laptop or desktop, capture or VNC, audio routing
Scenes, overlays, multiple cameras, chat integration, recording
Extra machine in the shack, audio and video sync work, heavier setup
Phone camera pointed at the screen
Phone and tripod
Zero configuration, works with any radio
Glare, moire, unreadable text, room noise instead of receiver audio
VNC or browser plus PC encoder
sBitx web interface or RealVNC, PC running OBS
Flexible layouts, remote operation doubles as the video source
Two machines, network loop risks, latency between control and stream
Public OpenWebRX receiver page
Hosted receiver
Audience tunes themselves, no streaming at all
No narration context, no digital mode app windows, depends on hosted hardware
For a club demo or a casual operating night, Pi Streamer wins on simplicity. For a produced net with lower thirds, multiple cameras and call in guests, graduate to OBS and use the Pi stream as one input among several.
Practical Uses for Clubs, Events and Field Stations
Live operating demonstrations. Run the HF station on the projector and on YouTube simultaneously so remote members follow along.
License class sessions. Show a real FT8 exchange, a CW filter narrowing onto a weak signal, or a logger entry being created, all live.
Special event stations. Give the event a public face beyond the log upload, with the station callsign and schedule in the stream description.
Workshop troubleshooting. Share the screen with a distant expert who can see the exact settings, error messages and waterfall behavior.
Antenna and filter comparisons. Switch antennas or adjust the audio tracking filter live while viewers hear the difference in the receiver audio.
Firmware and software demos. Volunteer builds from the sBitx community and official documentation updates are easier to assess when the audience watches them run.
Field debriefs. Record the activation stream and reuse segments as training material for the next portable outing.
One etiquette reminder: announce when a session is streaming, keep third party callsigns and personal data in mind before lingering on logger windows, and prefer a short delay if the stream doubles as contest or DX chase coverage.
Honest Limitations
sBitx centric testing. The package targets the sBitx environment. Other Pi applications are plausible but not individually verified, and audio routing is the usual stumbling block.
Shared CPU budget. Encoding competes with SDR processing on the same Pi 4. Aggressive 1080p settings can make the radio itself feel sluggish, which is worse than a soft looking stream.
Network dependence. No uplink, no broadcast. Field sites with marginal cellular hotspots should rehearse at 480p and keep expectations modest.
Stream key hygiene. A leaked key lets anyone broadcast to the channel. Rotate keys after shared or public configuration sessions.
YouTube onboarding friction. First time live accounts face verification and roughly a 24 hour wait. Third party restreaming or embedding policie

https://hamradio.my/pi-streamer-sbitx-youtube-live-raspberry-pi/

#streamer #raspberry #sbitx #ffmpeg #zip

#rss #feeds #linksharing #qwibbit

GrapheneOS Explained: The Private and Secure Mobile OS With ...

2026-11-05T06:14:58Z

GrapheneOS Explained: The Private and Secure Mobile OS With Android App Compatibility

Your phone knows where you sleep, who you call, what you photograph, which repeater you keyed up last night, and which hospital you visited last month. On a stock phone that knowledge is shared by default with the OS vendor, the app vendors, the ad networks they bundle, and anyone who can buy the resulting data. Most people accept this as the price of a working phone. GrapheneOS exists because a small group of researchers refused to accept that trade.
GrapheneOS at https://grapheneos.org/ is the project this article unpacks end to end: what it is under the hood, where it came from, what it feels like to live on every day, who gains the most from switching, who should probably wait, how to install it without bricking a Pixel, how amateur radio operators can use it in the field, and, frankly, what law enforcement officers and lawyers should understand about its encryption and seizure implications. No hype, no fear, just the technical reality as documented on the official site, in the source code, and in the project’s extensive FAQ and features pages.
Last updated: November 2026. Covers official device support from Pixel 6 through Pixel 10a, 7 year update guarantees on Pixel 8 and later, and current features documented at grapheneos.org.
TL;DR: GrapheneOS is a non-profit open source mobile OS based on the Android Open Source Project that runs on Google Pixel phones with locked verified boot. It hardens the OS against exploitation with a hardened allocator, hardened kernel, hardened browser, and rapid security updates, adds practical privacy controls such as a Network permission, Sensors permission, Storage Scopes, Contact Scopes, USB-C port control, auto-reboot, and a duress wipe PIN, and runs official Google Play as fully sandboxed regular apps with no special privileges. It suits journalists, activists, lawyers, clinicians, system administrators, business owners, and any everyday user who wants Android apps without OS level Google integration, including amateur radio operators who want a clean field phone. It isn’t for people who need a non-Pixel phone, carrier locked devices, or guaranteed compatibility with every banking app and game. Law enforcement should know it uses strong filesystem encryption with Titan M2 backed throttling, reboots return data to rest, a duress PIN can irreversibly wipe the device, and there’s no backdoor or vendor key to compel.
GrapheneOS homepage at grapheneos.org: the private and secure mobile OS with Android app compatibility, developed as a non-profit open source project. Screenshot by 9M2PJU.
What GrapheneOS Is
GrapheneOS is a privacy and security focused mobile operating system with Android app compatibility, developed as a non-profit open source project, that runs on Google Pixel phones with locked verified boot and ships no Google apps or services in the OS itself.
Unpack that definition and five properties fall out. First, it starts from the Android Open Source Project baseline, so regular Android apps install and run, including most apps that depend on Google Play. Second, it is hardened from the bottom up: sandboxing, exploit mitigations, and the permission model are all strengthened rather than bolted on. Third, privacy is structural, not a settings theme: there are no Google apps or services in the base OS, no analytics or telemetry phoning home, and default connections go to GrapheneOS servers instead of Google where a connection is needed at all. Fourth, it only supports hardware that can actually deliver on those promises, which in practice means recent Pixels with a secure element, verified boot with rollback protection, and long firmware support. Fifth, it’s a non-profit open source project with public source code, public releases, and a donation funded model, not a venture backed phone company.
The official site states the philosophy plainly on the features overview: substance over branding, exploit resistance against real adversaries, usability preserved where possible, and no bundling of incomplete third party Google service reimplementations such as microG. Users who want Google services can install the real thing inside the sandbox, covered in detail below. That single design decision explains most of what makes GrapheneOS different from every other custom Android distribution.
Where GrapheneOS Came From
GrapheneOS was founded by Daniel Micay in late 2014 as a solo research project. The early work was deeply technical and unglamorous: a port of OpenBSD malloc to Android’s Bionic libc, a port of PaX kernel hardening patches to supported kernels, and compiler toolchain hardening. Much of that era’s output landed upstream in AOSP, the Linux kernel, and LLVM, or heavily influenced the upstream fixes that replaced it. The full timeline is on the project’s history page.
In late 2015 a company was incorporated to sponsor the work through support contracts and custom variants. While under that sponsorship the project was known as CopperheadOS. The relationship collapsed in 2018 when the CEO attempted to seize control of the project, seized infrastructure, and took donations. The project walked away, rebranded first to AndroidHardening and then to GrapheneOS, and continued as a fully independent open source effort. The former sponsor has no association with the project today. Readers researching old CopperheadOS references should treat anything published by that company after 2018 as hostile to the project, which the history page documents directly.
Since the split, the project has grown to multiple full time and part time developers funded by donations, plus collaborating companies. In March 2023 the GrapheneOS Foundation was created as a Canadian non-profit to handle donation intake and distribution. That matters for trust: there’s no acquisition exit, no ad business, and no enterprise upsell shaping the roadmap. The donate page and public articles record show where money and effort go.
How GrapheneOS Protects You
GrapheneOS organizes its defenses in layers: shrink what attackers can reach, make the remaining bugs harder to exploit, contain breaches that still happen, and make persistence and data extraction expensive. What follows is a plain language tour of the mechanisms documented on the features page.
Attack surface reduction
The cheapest vulnerability is the one that can’t be reached. GrapheneOS strips unnecessary code, disables optional radios by default, and gates risky interfaces behind explicit user action. NFC, Bluetooth, and ultra wideband stay off until needed. Native debugging with ptrace is blocked for bundled apps. Most visibly, the USB-C port and pogo pins control blocks new USB connections as soon as the device locks, at both the hardware USB controller level and the kernel level, including USB-C alternate modes such as DisplayPort. The default mode is charging only when locked, and a full Off mode also disables charging protocol attack surface. This single toggle defeats a whole class of locked device forensic tools that depend on USB.
Exploit mitigations that matter
Most remote and local code execution starts as memory corruption in unsafe languages. GrapheneOS answers with its own hardened_malloc, a hardened Bionic libc, a hardened kernel with slab canaries plus zero on free, four level page tables with high entropy ASLR on arm64, and hardware memory tagging on ARMv9 Pixels where the silicon supports it. Android Runtime JIT profiling is replaced with full ahead of time compilation, dynamic code loading is blocked for nearly the entire base OS, and user installed apps get optional toggles to block dynamic code loading from memory and storage plus WebView JIT. On ARMv9, branch target identification, pointer authentication, and shadow call stack coverage go beyond stock Android. For readers who don’t follow allocator internals, the practical summary is simple: exploit chains that work reliably against stock firmware tend to break, crash, or become unreliable on GrapheneOS, which is exactly the point.
Sandboxing and verified boot
The Android app sandbox, the browser renderer sandbox, media codec sandboxes, and user profiles all sit on top of a hardened SELinux and seccomp policy plus the hardened kernel underneath. Verified boot is strengthened with continuous fs-verity enforcement for out of band system app updates, downgrade protection, and disabled compressed APEX to cut boot chain attack surface. An enhanced hardware attestation story through the Auditor app lets a second device verify the OS running on the first, which is rare in the mobile world and directly useful for high risk users who need to confirm a phone hasn’t been tampered with.
Patching speed
GrapheneOS ships full monthly Android Security Bulletin patches plus its own fixes, and on GKI devices it tracks the latest Linux kernel LTS point releases rather than waiting for quarterly freezes. At various points this has put GrapheneOS months ahead of stock Pixel firmware on kernel patch level, worth hundreds of upstream fixes. The project also reports vulnerabilities upstream itself, including Pixel specific issues and missed backports it discovers during its own integration work.
Apps and services built by the project
The OS ships a small set of purpose built apps: Vanadium, a hardened Chromium based browser and WebView with per site JIT control; a minimal security focused PDF Viewer; a modern Camera app; the Auditor attestation app; a Setup Wizard; Seedvault encrypted backups; and a GrapheneOS app repository client that updates bundled apps out of band. The usage guide documents each one. Notably absent by design: anything from Google, any bundled app store with ads, and any third party app choices imposed on the user.
Privacy Controls You Can Actually Touch
Security keeps attackers out. Privacy decides what legitimate apps are allowed to see. GrapheneOS adds a set of user facing controls that stock Android lacks.
Network permission toggle. Deny an app all network access, direct and indirect, including localhost tricks used to bridge profiles. The OS reports the network as down so apps degrade gracefully instead of crashing.
Sensors permission toggle. Block accelerometer, gyroscope, compass, barometer, thermometer, and other sensors not covered by Camera, Microphone, or Activity Recognition permissions. Apps receive zeroed data, and the OS posts a notification so the user knows which app asked.
Storage Scopes. Instead of handing an app your whole photo library, let it believe it has storage permission while it only sees its own files plus the specific folders you add. This fixes the all or nothing storage dilemma that pushes users to over share.
Contact Scopes. Same idea for contacts: present an empty address book by default and grant access to selected contacts or groups only.
Secure Paste. Replaces the global clipboard free for all with per app control. Apps can’t silently sip clipboard content set by other apps.
USB-C port and pogo pins control. Covered above, but it is as much a privacy control as a security one: no data leaves over a cable you didn’t authorize.
Auto-reboot. After a configurable 10 minutes to 72 hours of lock time, default 18 hours, the phone reboots itself so encryption keys leave memory and data returns to rest. Implemented in init so a system process crash can’t suppress it.
Duress PIN and password. A separate credential that irreversibly wipes the device including installed eSIMs the moment it is entered at any unlock prompt. If the duress credential ever equals the real one, the real unlock wins so accidents can’t wipe the phone.
Two factor fingerprint unlock. Fingerprint plus a short PIN completes unlock, pairing a strong primary passphrase with convenient secondary unlock. Failed PIN attempts count toward the standard limit.
Longer passwords and PIN scrambling. Up to 128 character passwords by default, plus optional scrambled PIN pad layout against shoulder surfing and side channels.
Hardened user profiles. Up to 32 profiles, per profile encryption keys, end session buttons that purge keys without a full reboot, per user app install controls, and notification forwarding controls. Profiles are the single most powerful privacy tool in the OS.
Private screenshots. Strips EXIF build fingerprint and local time metadata from screenshots by default, with an opt in toggle for users who want timestamps back.
Network location choice. Opt in local position estimation with a GrapheneOS proxy to Apple data today and a self built database in progress, with in memory caching that keeps working briefly offline. No Google network location in the base OS.
Wi-Fi privacy and LTE-only mode. Per connection MAC randomization with DHCP state flush, IPv6 privacy fixes backported where needed, and an LTE-only mode that disables 2G, 3G, and 5G radio code to shrink cellular baseband attack surface.
Privacy by default. No Google in the base OS, personalized keyboard learning off, sensitive lock screen notifications hidden, passwords hidden during entry, and default service endpoints pointed at GrapheneOS for connectivity checks, time, PSDS satellite data, SUPL, and remote provisioning, each with toggles to use Google or disable where applicable.
Together these controls change the daily question from do I trust this app into what exactly does this app get. That shift is the whole appeal for privacy conscious users.
Sandboxed Google Play: How Android Apps Still Work
The most misunderstood GrapheneOS feature is also the reason most people can switch without giving up their apps. GrapheneOS will never ship Google Play services in the OS and will never bundle microG. Instead it offers sandboxed Google Play: the official Google Play Store, Play services, and Services Framework installed as ordinary apps inside one user profile, with zero special privileges, no system backend role, and no access to other profiles.
Because Play runs as a regular app, it can only see what the user grants it inside its own profile, and only apps in the same profile that explicitly choose to call it can use it. Nearly everything works: Play Store purchases and license checks, in app billing, Play Asset and Feature Delivery, Firebase Cloud Messaging push, dynamically downloaded modules, SafetyNet and Play Integrity where the app doesn’t insist on a Google certified OS verdict, and rerouted or native network location at the user’s choice. Android Auto is supported through an extension of the same compatibility layer with reduced privileges, documented in the usage guide.
What doesn’t work is anything that inherently requires OS privileges: device admin style remote control, privileged background powers stock firmware grants Play, and apps that hard refuse to run unless the OS passes Google certification. In practice the working set covers banking apps, transit apps, messaging apps, and games for most users, but outliers exist per bank and per country, so testers should install their must have apps in a secondary profile during week one and confirm push notifications, tap to pay, and QR login flows before committing.
Who Should Use GrapheneOS
Short answer: anyone with a supported Pixel who wants Android apps without OS level Google integration and is willing to spend an afternoon learning profiles. Longer answer by role:
Journalists and editors handling sensitive sources. Profiles separate source communications from personal life, auto-reboot bounds seizure exposure, Auditor verifies device integrity before travel, and Vanadium with JIT disabled by default shrinks drive by web risk.
Activists, organizers, and human rights workers. Same toolkit plus duress PIN planning, per connection Wi-Fi randomization at venues, and LTE-only mode in hostile radio environments.
Lawyers, clinicians, and fiduciaries. Contact Scopes and Storage Scopes enforce least privilege on client and patient data, encrypted Seedvault backups satisfy continuity needs, and no OS level ad telemetry simplifies confidentiality duties.
System administrators and security engineers. Fast patching, memory tagging,exec spawning protections, and a readable source tree make it the rare phone that rewards inspection instead of punishing it.
Business owners and finance staff. A dedicated Owner profile plus a banking profile with sandboxed Play isolates money apps from games and social a

https://hamradio.my/grapheneos-privacy-os-secure-android-guide/

#securemobile #androidapp #pixel #googlepixel #appcompatibility #opensource #sourceproject #scopes #contactscopes #storagescopes

#rss #feeds #linksharing #qwibbit

PCBView.io: Open Any PCB Design in Your Browser with Nothing ...

2026-11-04T20:17:42Z

PCBView.io: Open Any PCB Design in Your Browser with Nothing Uploaded

Every homebrewer knows the moment. Someone drops a link to an open source radio board, a filter, an antenna tuner, a QRP transceiver, and the repository holds a zip of Gerbers with cryptic Protel extensions. To actually see the board you’re expected to install a full EDA suite, learn its importer quirks, and wait through a splash screen, all for a two minute question: does this look right, will it fit my enclosure, is that the footprint I think it is. Club mailing lists quietly fill with screenshots taken on someone else’s machine because nobody wants to be the one who installs 4 GB of CAD software to answer a yes or no question.
PCBView.io removes that whole ritual. It’s a free viewer at https://pcbview.io that opens just about any PCB design file directly in the browser: Gerber RS-274X with X2 and X3 attributes, KiCad boards, Altium PcbDoc, Cadence Allegro and OrCAD .brd, Eagle XML, EasyEDA Standard and Pro, ODB++ jobs, PADS ASCII and binary, IPC-2581, and DipTrace, plus Excellon drills, pick and place, BOMs, IPC-D-356 netlists, and plain archives. Everything is parsed locally on your own computer and nothing is uploaded to any server, which is stated twice in the interface itself, once on the landing card and once in the status line after every import. Version 1.0.2 at the time of writing is the work of Derivative Engineering LLC, contactable at ben@pcbview.io, and it was introduced to the wider electronics community through an EEVblog forum thread by its developer.
Last updated: November 2026. Current release v1.0.2, published 4 November 2026.
TL;DR: PCBView.io is a free browser based PCB design viewer by Derivative Engineering LLC that opens Gerber (X2/X3), KiCad, Altium, Allegro, Eagle, EasyEDA, ODB++, PADS, IPC-2581 and DipTrace files locally in your browser with nothing uploaded. Drop files, a folder, or a GitHub link, or open the built-in BeagleBone AI example, then inspect layers, search parts and nets, read silkscreen text, flip the board, switch mm and mil, render a 3D view, and export PNG images, mechanical STEP models, or shareable board zips. A matching desktop app for Windows, macOS and Linux runs fully offline. It is free to use with optional pay-what-you-like support through Stripe. It isn’t open source software: the web bundle is proprietary to Derivative Engineering LLC.
PCBView.io with the built-in BeagleBone AI example design: 23 layers, 716 parts, 595 nets, all parsed locally with nothing uploaded. Screenshot by 9M2PJU.
What PCBView.io Is
PCBView.io is a free browser based PCB design viewer that opens Gerber, KiCad, Altium, Allegro, Eagle, EasyEDA, ODB++, PADS, IPC-2581 and DipTrace board files locally in your computer’s browser, with no uploads, no account, and no EDA installation.
Unpack that sentence and the design philosophy becomes clear. The landing page is a single card that says “import a pcb” with exactly three ways in: open files, open a folder, or paste a GitHub link, plus a one-click example design for anyone who arrived empty handed. Below the buttons sits a row of format pills, Gerber through DipTrace, and a link to the full format list and help. There’s no login wall, no project dashboard, no cloud workspace trying to upsell you. The viewer parses the copper, mask, paste, silkscreen, drills, parts, and nets on your own machine, draws the board in 2D and 3D, and keeps your color choices in the browser. When you want to share what you opened, the app zips the files you supplied into one archive that someone else drops onto the same page. At no point does a board file travel through a company server, which is the feature that makes this tool usable for unreleased commercial work and club designs under NDA as well as open hardware.
Three facts define the project. First, it is format coverage first: ten native board formats plus drills, placement, BOM, netlist, and nested archives, with per-format notes about exactly which versions and quirks are handled. Second, it is privacy architecture rather than privacy policy: the files never leave because the parser runs client side, so there’s no server side promise to audit. Third, it’s a small commercial freeware project, not a community open source one: free to use, pay what you like toward development through Stripe, copyright 2026 Derivative Engineering LLC, with the JavaScript bundle explicitly marked proprietary. That distinction matters and this article states it plainly, because many viewers in this space are genuinely MIT licensed while this one isn’t.
Why a Browser Viewer With No Uploads Matters
To appreciate PCBView.io, start with how board review usually works. The established paths are a desktop Gerber viewer such as Gerbv or KiCad’s GerbView, a manufacturer portal uploader such as a fab’s online quote page, or a full CAD seat. Each path has a cost. Desktop tools need installing and updating per operating system and often understand Gerbers but not native PcbDoc or .brd files. Manufacturer portals understand plenty but require uploading your intellectual property to a third party server, which is unacceptable for client work, contest designs, or anything under NDA. Full CAD seats understand everything but weigh gigabytes and assume you already speak that tool’s dialect.
PCBView.io collapses all three paths into a URL. A club Elmer can review a member’s first KiCad board without installing KiCad. A buyer can check an ODB++ job from a contractor on a Chromebook. A ham staring at a graduated engineer license exam project or a homebrew transceiver fork on GitHub can paste the repo link and see the board in seconds. The GitHub path deserves emphasis: paste a repo, folder, or file URL and the viewer finds the boards inside it, design files, ODB++ jobs, folders of Gerbers, fab zips, and lets you pick which to open, with the files fetched straight from GitHub to the browser. Public repositories only, which is the correct scope for a client side tool with no credentials.
The privacy angle is the load bearing one. Tracespace, the well known open source Gerber renderer behind many portals, also parses client side, and PCBView.io shares that architecture without sharing its license. Nothing uploaded means the tool works identically for an open hardware release and a pre-patent prototype. It also means the viewer keeps working on flaky field internet once loaded, and the desktop builds take the idea to its conclusion by running with the network unplugged entirely.
Every Format It Opens
The format list below is taken from the viewer’s own help text, which is refreshingly specific about versions and rendering compromises instead of claiming universal support.
Format
Files accepted
What the viewer reads
Gerber
.gbr, .gtl, .gbl and other Protel names, or any extension
RS-274X, X2 and X3 from any EDA tool. X2 attributes add nets, pins, and parts so copper becomes searchable. X3 component files add position, rotation, value, part number, and footprint per part. Layers are recognized from X2 file functions or file names.
KiCad
.kicad_pcb
Board files from KiCad 4 through 10, with parts, values, footprints, nets, pours, per-layer pad stacks, and mask margins.
Altium
.PcbDoc
Altium Designer boards: copper, pours, mask, paste, silkscreen with text, mechanical layers, and drills, with parts, parameters, part numbers, and nets. TrueType text is drawn in a stroke font. Parts marked not fitted in a project variant show as fitted, since variants live in the project file.
Allegro
.brd
Cadence Allegro and OrCAD PCB Editor boards from Allegro 16 onward, checked against boards saved by 16.3 through 17.5. Each layer is drawn from the board’s own artwork films, showing what Allegro plots: copper with pour fills and negative planes, mask, paste, silkscreen, drawing layers, and drills. Boards saved by Allegro 15 or older need re-saving in a newer Allegro.
Eagle
.brd
XML boards from Eagle 6 onward. Eagle stores pours as outlines, so they are filled here by approximation.
EasyEDA
.json (Standard), .epro, .epro2, .eprj, project folders (Pro)
EasyEDA Standard and Pro, also known as LCEDA and JLCEDA. Pro 2 and 3 project files from the online editor and desktop client. Reads parts with LCSC and manufacturer part numbers, nets, pours, text, and imported artwork. A loose Pro .epcb board has no footprints, so its parts show without pads.
ODB++
Job folder, .tgz, .zip
ODB++ v7 and v8 jobs from any EDA tool, tested with Altium, Cadence Allegro, Siemens Xpedition and PADS, KiCad, Autodesk Fusion, EasyEDA, and DipTrace. Reads layers, parts with values and part numbers, nets, and drill holes.
PADS
.asc, .pcb
PADS Layout and PowerPCB: ASCII .asc from PowerPCB 5 through PADS VX, plus binary .pcb. Reads layers, parts, footprints, nets, tracks, vias, saved pour fills, and text. Part values and attributes come from .asc files.
IPC-2581
.xml, .cvg
Revisions A through C. One file holds the whole board including its BOM.
DipTrace
.dip
DipTrace boards from DipTrace 2 through 5 (file formats 37 through 62). Reads layers, parts with values, footprints, nets, tracks, vias, saved pour fills, board text, shapes, and part markings. Silkscreen pictures are not drawn.
Excellon drills
.drl, .xln, .txt and others
Drill and route files, plated and non-plated. Altium .txt drills are recognized by content.
Pick and place
.csv, .pos, .txt, .mnt and others
Placement files from KiCad, Altium, Eagle, and generic CSV. Units and columns are worked out from the file.
BOM
.csv, .tsv, .txt
Comma, tab, or semicolon separated. Matched to parts by reference designator, so search finds values and part numbers.
IPC-D-356 netlist
.ipc, .d356, .356
Bare-board test netlist: net names for pins and test points.
Archives
.zip, .tgz, .tar.gz, .tar
Opened in place, with nested archives and folders included.
Two honest limitations are documented inline and worth repeating. Drill chart symbols on Allegro drawing layers aren’t drawn, and DipTrace silkscreen pictures are skipped. Eagle and EasyEDA pours saved without their fill are approximated rather than exact. These are the normal scars of supporting eleven layout tools without owning any of them, and the viewer discloses each one at the point of use instead of failing silently.
Tour of the Viewer: From Drop to Net Highlight
Opening a board takes seconds. Drop files or a folder onto the page, or click through the buttons, and a status sequence narrates the import: reading files, unpacking archives, identifying board files, merging layers, parts and nets, then drawing the board, with an elapsed-seconds readout in the footer when it finishes. The built-in example is the BeagleBone AI hardware, loaded from its Eagle .brd on GitHub, which opens as a 23-layer, 716-part, 595-net Allegro-class demonstration that immediately shows what the layer panel and search box are for.
The 2D view is the working surface. Toggle between top and bottom copper with one click, mirror the view as if the board were turned over, flip between 2D and flat-shaded 3D, and show or hide reference designator labels. Units switch between millimeters and mils. Each layer gets its own row with color chip and checkbox, double-click or press and hold a layer to isolate it, and plated and non-plated holes get independent toggles. The viewer remembers a sensible default of top copper, top silkscreen, holes, and outline, and a reset control restores it.
Search is the standout interaction. One box covers parts, values, part numbers, nets, and pins, with an optional pass that reads part designators straight from the silkscreen artwork using on-device text recognition. Pick a net and the viewer highlights its copper, lets you widen the selection from the object under the cursor to everything connected on that layer and then on all layers, and reports routed track length per net. If the imported files carry no connectivity, the panel says so directly and tells you what would fix it: Gerber X2 with net attributes, an IPC-D-356 netlist, or a native design file.
Solder mask, silkscreen, and finish colors are adjustable, including ENIG-style gold versus HASL, silver, tin, bare copper, and OSP treatments plus arbitrary mask and silk colors, with a reset to defaults. Colors persist in the browser across boards. Keyboard, mouse, and touch operation are all first class: scroll to zoom, drag to pan, tap to select copper or a part under the finger, tap again to widen, two-finger tap and press-and-hold variants on touch, and a hold-to-preview gesture that shows what sits under the finger without committing to a selection.
3D View, STEP Export, Images, and Sharing
The 3D button renders the board with its solder mask, silkscreen, and finish, extruded to the board thickness with cutouts, slots, and mounting holes. Navigation is orbit style with fit and reset controls. It’s a presentation view rather than a mechanical CAD model, but the export next to it crosses that gap: Save STEP writes a small 3D model for mechanical CAD containing the board at its thickness with cutouts, slots, and mounting holes, plus each part as a box named by its designator. Vias and small pins are omitted, identical parts share one shape definition, and the file never exceeds 5 MB. That is exactly the envelope a 3D-printed enclosure designer or a front-panel layout needs: true outline, true holes, approximate components.
Save image exports a PNG of the current view, which is how the screenshots in this article were produced and how club newsletters get their board illustrations. Save board files collects the files you opened into one zip so the next person drops it onto PCBView.io and sees the same board, with the help text making the privacy point a third time: whoever gets it opens it locally, since nothing goes through a server. Together the three exports cover the three audiences for board review: the forum post (PNG), the enclosure designer (STEP), and the collaborator (zip).
Desktop App: The Same Viewer With the Network Unplugged
PCBView.io also ships as a native-feeling desktop app for Windows, macOS, and Linux, served from the pcbview-releases GitHub repository. The file list is short and conventional: a Windows installer, separate Apple silicon and Intel disk images for macOS, and a Linux AppImage, with Open With integration so Gerbers and board files launch straight into the viewer. Functionally it’s the same viewer, with one stated difference: it runs with the network unplugged, and the only thing it ever asks the internet for is whether a newer version exists.
That offline property is the point. Field laptops, contest stations, and EMC test chambers with no connectivity get the full viewer. The one platform quirk is documented in the help: on macOS the first launch needs the familiar System Settings, Privacy and Security, Open Anyway approval for an unnotarized developer build. Automatic update checks keep the app current without an app store, and the web version needs no install at all.
PCBView.io Against the Usual Viewers
No tool covers every workflow, so here is where PCBView.io sits relative to the viewers hams and homebrewers actually use.
Viewer
Runs
Native board files
Gerber + drills
3D / STEP
Privacy model
License and cost
PCBView.io
Browser, plus offline desktop app
KiCad, Altium, Allegro, Eagle, EasyEDA, ODB++, PADS, IPC-2581, DipTrace
Yes, X2/X3 with nets and parts
3D view plus STEP export under 5 MB
Local parsing, nothing uploaded
Proprietary freeware, free to use, pay what you like
KiCad GerbView
Desktop
KiCad native, Gerbers only otherwise
Yes
No 3D board assembly view
Fully local
Open source, free
Gerbv
Desktop
No, Gerber and drill only
Yes, RS-274X and Excellon
No
Fully local
Open source, free
Tracespace
Browser
No, Gerber focused
Yes, high fidelity SVG render
No
Client side render, nothing uploaded
MIT open source, free
Manufacturer portal viewers
Browser
Varies by fab
Yes
Usually 3D render only
Files uploaded to vendor servers
Free with quote flow
Altium 365 viewer
Browser
Altium native, best in class
Yes
Yes, full MCAD co-design
Cloud workspace, account required
Commercial, free tier limited
Read the table as a decision guide. If you need a guaranteed-correct Altium collaboration space with commenting and supply chain data, the vendor cloud still wins. If you need an embeddable MIT licensed renderer inside your own tooling, Tracespace rem

https://hamradio.my/pcbview-io-browser-pcb-design-viewer/

#pcbdesign #nothinguploaded #derivativeengineering #browser #open #engineeringllc #odb #gerber #altium #eagle

#rss #feeds #linksharing #qwibbit

DroidStar VU3LVO for iOS: The iPhone Digital Voice Client for ...

2026-11-04T16:48:41Z

DroidStar VU3LVO for iOS: The iPhone Digital Voice Client for DMR, M17, YSF, P25 and NXDN

There is a particular frustration every digital voice operator knows. The talkgroup is alive, the net is starting in five minutes, and the only computer in your pocket is an iPhone. Your DMR handheld is on the charger at home, the hotspot is in a drawer, and the App Store has never had a serious native client for BrandMeister. For years, iPhone owners watched Android users work the world on DroidStar while they stayed on the sidelines.
DroidStar VU3LVO iOS ends that wait. It’s a free, open source digital voice client for iPhone and iPad, built by Rohith Namboothiri VU3LVO of Bengaluru, India, that puts DMR on BrandMeister with the modern ODTP protocol, plus YSF, M17, P25, NXDN, D-STAR and AllStar, all in a native-feeling iOS app with live transcription, a Talk Group scanner, a persistent QSO log, and Dynamic Island status. It is used by more than 10,000 hams worldwide, it costs nothing, and it needs no radio and no hotspot, just a licence, a DMR ID, and an internet connection.
Last updated: November 2026. Current release v0.45.04 with v0.45.05 TX audio pass rolling out via TestFlight.
TL;DR: DroidStar VU3LVO iOS is a free GPL-3.0 iPhone and iPad digital voice client for licensed hams, forked from Doug McLain AD8DP DroidStar and rebuilt for iOS by Rohith VU3LVO. It works DMR over BrandMeister using ODTP with automatic fallback, plus YSF, M17, P25, NXDN, D-STAR and AllStar over the internet with no radio or hotspot. Headline features are on-device live transcription, Dynamic Island and Lock Screen status, an HD Voice TX suite with up to 24 dB pre-amp boost, a Talk Group scanner with favourites, net reminders, and a persistent ADIF and CSV QSO log. Distribution is Apple TestFlight invite only because of the 10,000 tester cap. Setup takes a callsign, a free RadioID DMR ID, and a BrandMeister SelfCare password.
DroidStar DMR for iOS by VU3LVO. Your iPhone on the air with live transcription, Talk Group scanner, and HD Voice TX. Screenshot: vu3lvo.com/droidstar.
What DroidStar VU3LVO for iOS Is
DroidStar VU3LVO iOS is a free, open source digital amateur radio client for iPhone and iPad that connects a licensed operator directly to DMR, YSF, M17, P25, NXDN, D-STAR and AllStar networks over the internet, with no radio, no hotspot, and no external vocoder hardware.
Unpack that sentence and the whole project comes into focus. The app is a modified fork of the original DroidStar by Doug McLain AD8DP, rebuilt by VU3LVO as a full iOS experience on top of the same proven Qt and C++ digital voice core. The phone behaves like a hotspot and a transceiver at once. Your voice is captured on the iPhone microphone, cleaned up with noise cancellation and levelling, encoded on-device into the compact digital stream each mode expects, and sent over Wi-Fi or cellular to a BrandMeister server, a YSF reflector, an M17 reflector, or an AllStar node. Replies come back the same way and play through the speaker, headphones, or a Bluetooth headset. The RF path is replaced by the internet path, and everything else, talkgroups, reflectors, callsign routing, QSO logging, works the way radio operators already understand.
Three facts define the project. First, it is iPhone first. Android development is paused while every hour goes into the iPhone and iPad app, with a rebuilt Android app and a browser based DroidStar for Web on the public roadmap. Second, it is genuinely open source under GPL-3.0, with the full source at github.com/rohithzmoi/Droidstar-DMR, currently around 77 stars and 20 forks. Third, it is free in both senses. No ads, no subscriptions, no premium tier. Community donations through Buy Me a Coffee and PayPal pay the Apple developer fees and test hardware, and all 157 contributors are thanked by name on the project contributors wall.
Where It Came From: From dstar-monitor to a 10,000 Ham iPhone App
To understand why this fork exists, rewind to the beginning of the DroidStar story. Doug McLain AD8DP started with dstar-monitor, a receive only D-STAR listener that decoded traffic over the network using the MBElib library. When transmit support and software vocoding arrived, it became a true two way client, renamed from DudeStar to DroidStar, and shipped on Android through the Google Play Store plus Linux, Windows, macOS, and iOS builds from one Qt codebase. It spoke M17 as early as October 2020, one of the first clients to do so, and it grew into the most versatile digital voice app in the hobby: DMR, M17, YSF and FCS in DN and VW modes, P25 Phase 1, NXDN, D-STAR on REF, XRF and DCS reflectors, and AllStarLink as an IAX2 client, all over UDP, with support for USB AMBE vocoders such as ThumbDV and DVstick and for MMDVM modems in hotspot or direct mode.
Stock DroidStar, for all its power, always felt like a developer tool that happened to run on a phone. Configuration meant hunting for a vocoder plugin URL and pasting it into Settings, hoping the host was still online. The interface was a flat stack of Qt dropdowns. On iOS it existed but never felt at home.
Rohith VU3LVO forked the project in August 2024 to fix exactly that. The first public builds landed on Android and iOS together, and the theme was immediate: operator name and country looked up automatically from the DMR ID and shown under every callsign, a QSO tab logging the most recent 250 contacts with callsign, DMR ID, talkgroup, handle, country, and local time, persistent logs that survived restarts, ADIF and CSV export with per-entry selection and share sheets, one tap links from any log row to the station QRZ, BrandMeister, and APRS pages, a recent talkgroup dropdown so nobody had to memorise numbers, Last Heard on the home page, and background receive so looking up a station on QRZ didn’t kill the audio. September 2024 added shareable log files, a Chinese build translated by BD4VHI, full iOS 18 support, and external earphone buttons as PTT on Android. Then came the hard call documented in the repository announcement: maintaining both platforms at solo-developer quality wasn’t sustainable, so Android was frozen and every hour moved to the iPhone and iPad app.
January 2026 brought the big redesign: a completely new interface with Dynamic Island, Lock Screen live activities, Notification Center status, and control over whether your own transmissions are logged. October 2026 delivered the AI wave across four rapid releases, v0.45.01 with beta live transcription and net reminders, v0.45.02 with the HD Voice suite, favourite talkgroups, and the TG scanner, v0.45.03 with the pre-amp volume boost up to plus 24 dB, and v0.45.04 with ODTP for BrandMeister, IAX WT, AllStar passwords, and a rebuilt TX audio pipeline. Version 0.45.05 followed as a focused audio pass that lifts the muffled close-talk quality out of transmit audio before encoding. Each step was shaped by tester feedback, including ODTP itself, which was built because a tester asked for it and explained why.
Today the numbers tell the story: more than 10,000 licensed hams on the air, callsigns from 18 countries on the contributors wall, 183 community contributions, and active discussion on Reddit communities, a WhatsApp channel for polls and sneak peeks, and GitHub issues.
How It Works: Your iPhone Becomes the Radio
The signal path has four stages, and the project page explains them in plain language worth repeating here.
First, you press PTT. DroidStar records through the iPhone or headset microphone and cleans the audio with noise cancellation, the HD Voice enhancer, and the mic leveller before anything is encoded. Hold the on-screen PTT or hold Space on a connected keyboard.
Second, the audio is encoded on-device. The built-in vocoder turns your voice into the same compact digital stream a hardware radio would produce for the selected mode. No external dongle, no plugin download, no AMBE USB stick required.
Third, the stream goes out over the internet. Over Wi-Fi or cellular, the app talks to a BrandMeister server using ODTP, or to a YSF, M17, D-STAR, or AllStar reflector using that network native protocol.
Fourth, the network carries you worldwide. Your transmission goes out to linked repeaters, hotspots, and apps on the selected talkgroup, and replies return the same way in under a second.
What you need on your side is small: a valid amateur licence and callsign, a free numeric DMR ID from RadioID.net, and for BrandMeister DMR, a SelfCare password from the BrandMeister portal. No licence means no invite, the TestFlight form requires you to confirm you hold one, and the app is explicitly beta software for licensed hams only.
ODTP for BrandMeister: The Headline Protocol Upgrade
The single most technical change in v0.45.04 is Open DMR Terminal Protocol, or ODTP, support for BrandMeister hosts, and it deserves a section of its own because it changes daily operating habits.
ODTP is a modern protocol for connecting DMR terminals and apps to BrandMeister, carried over UDP port 54006. In the older HomeBrew based workflow, changing talkgroups involved a key-up dance of connect, kerchunk, and confirm steps that every hotspot user knows. With ODTP, talkgroup changes happen cleanly without that ritual. The app prefers ODTP whenever the selected BrandMeister host offers it and falls back automatically to the standard protocol when it doesn’t, so a link never dies just because a server lags behind. A toggle in Settings turns ODTP off entirely for operators who want the legacy behaviour.
Practically, this means faster, more reliable BrandMeister sessions from a phone on hotel Wi-Fi or a moving train. The fallback design is the right call for a mobile app, because phone networks drop, change address, and roam between cells, and the last thing a portable operator needs is a protocol that punishes that. The project credits a tester suggestion for the feature, which fits the pattern of this fork: the most useful networking change in a year started as a user request.
Every Feature That Matters on the Air
Most of these features were requested by the people who use the app, and the project page makes them interactive so visitors can try the audio tools in the browser. Here is what each one does in practice.
Multilingual Live Transcription in Beta
Read incoming transmissions as text while they are still happening. Two engine choices cover two situations. Apple built-in speech recognition costs nothing extra and works well in quiet conditions. Four downloadable models run entirely on the phone for noisy bands or places where a speaker isn’t an option, such as a meeting room or a night train. Transcripts can sit alongside the QSO log. It’s still marked beta, and the developer is honest that a noisy marginal signal will still fool it. For net control, for operators with hearing loss, and for anyone monitoring a talkgroup with the volume down, this is the most accessibility positive feature any DroidStar variant has shipped.
Dynamic Island, Lock Screen, and Notification Center
TX and RX state, the active callsign, and the talkgroup stay visible without opening the app. When the phone lies face down on the bench and a callsign comes up, one glance answers the only question that matters. This was among the most requested features in the project history, and it’s easy to see why. A radio has a display you can read across the room. A phone app that hides everything behind a lock screen isn’t a radio replacement. This closes that gap.
HD Voice Audio Suite and 24 dB Pre-Amp
Three TX enhancements switch independently from the home page: hardware plus neural noise cancellation with a gate, a wideband presence and clarity enhancer for your own voice, and an automatic mic volume leveller. Above them sits a pre-amp volume boost adjustable up to plus 24 dB for situations where the phone microphone is simply too far away, such as a dashboard mount or a speakerphone style shack setup. The project page includes a noisy sample clip and a record-your-own-voice demo so you can flip each switch and hear the difference through headphones. Version 0.45.05 extended this work with a dedicated pass against muffled close-talk transmit audio. The fair way to judge any of it’s the Parrot test on talkgroup 9990, which plays your own transmission back exactly as the network heard it.
Talk Group Scanner and Favourites
Star the talkgroups you actually use, scan those plus currently busy global BrandMeister talkgroups, and tap once to tune in when a conversation you want catches your ear. This replaces the old workflow of typing numbers from memory or keeping a notes file of favourites. For travellers, it’s the fastest way to find activity on an unfamiliar repeater network.
Net Reminders
Save the schedules of your regular nets and get a push notification at check-in time. Anyone who has ever sworn they would remember the Tuesday net and then remembered on Wednesday understands why this exists.
Bluetooth, Wired Headsets, and Background Receive
Transmit can use a Bluetooth or wired headset microphone. Receive keeps playing when you leave the app and when the iPhone silent switch is engaged. The developer notes he uses this constantly, typically while looking up the station just heard on QRZ, and that is exactly the workflow it enables: monitor in the background, research in the foreground, tap back to answer.
Many Modes, One App
DMR on BrandMeister with ODTP, YSF rooms, M17 reflectors, P25, NXDN, D-STAR reflectors across REF, XRF, and DCS systems, and AllStar over IAX including IAX WT and password protected nodes. The full inherited DroidStar mode set is present. So one app covers the weekly DMR net, the YSF room your club uses, and the AllStar node that links your local repeater.
Recent Talk Groups and Last Heard
The TG field remembers what you have dialled, so 91, 3100, 235, and your local group behave like speed dial instead of phone numbers you are afraid of forgetting. Last Heard shows the two most recent stations on the home page, with the AMBE status notice moved to Settings where it belongs.
A QSO Log That Exports Anywhere
Logging is where the VU3LVO fork first distinguished itself, and it remains the best reason to try it beyond the iOS polish. Every station heard is logged automatically with callsign, DMR ID, talkgroup, operator handle, name, country, and local time. The log persists across restarts. You control how long it is allowed to grow and whether your own transmissions are logged at all.
Export covers the formats loggers actually accept. Select entries and export as ADIF for direct import into general loggers or as CSV for spreadsheets and analysis, then share by AirDrop, email, or WhatsApp. An illustrative sample from the project page shows the shape:
Call
DMR ID
TG
Name
Country
Local
VK3TV
5053241
91
Mal
AU
21:42
GW4VXE
2350912
235
Tim
GB
21:35
K7RAN
3153310
3100
–
US
21:28
DG9MAQ
2624456
262
–
DE
21:21
ZL2BEZ
5302019
91
–
NZ
21:14
One tap on any entry opens that operator QRZ, BrandMeister, or APRS page. The name and country under every callsign come from the DMR ID lookup, so the log reads like a well kept station diary instead of a bare number list.
Getting on the Air: From TestFlight Invite to First QSO
Distribution is the one part that surprises newcomers. There’s no App Store listing. Apple TestFlight caps each app at 10,000 testers, so invites go out one by one instead of through a public link.
Start at vu3lvo.com/droidstar/testflight and fill in the form with your full name, callsign, the email address tied to your Apple ID, your optional DMR ID, your country, your device type, and how you heard about the app. Confirm you hold a valid amateur licence. Install Apple free TestFlight app from the App Store on your iPhone or iPad. When the invite email arrives, open it, tap Start Testing, and install DroidStar. If anything about the invite looks wrong, write to hi@vu3lvo.com.
Before you transmit, secure the two credentials every digital voice app needs. Register your callsign at RadioID.net for a free numeric DMR ID, typically seven digits, and allow a day or two for new IDs to propagate across networks. For BrandMeister DMR, create an account on the BrandMeister SelfCare portal with your callsign and DMR ID and set a SelfCare password, which is what the app sends to authenticate, stored in the BM Pass field.
Then configure the app. Open Settings and enter your callsign, DMR ID, and BM

https://hamradio.my/vu3lvo-droidstar-ios-digital-voice-client/

#voiceclient #dmr #ios #talkgroup #nxdn #ysf #iphone #droidstar #dynamicisland #brandmeister

#rss #feeds #linksharing #qwibbit

FieldMesh: The Outdoor-Ready MeshCore Firmware with GPS Tracking, ...

2026-11-04T05:49:09Z

FieldMesh: The Outdoor-Ready MeshCore Firmware with GPS Tracking, Off-Grid Mode and SOS

There is a particular kind of radio trip every outdoor group knows. The hike spreads out along the trail, the festival camp splits across three stages, the 4×4 convoy stretches around two bends, and the phone network that was supposed to hold everyone together quietly dies. One bar, no bar, then that familiar ritual of holding the phone to the sky. Mesh radio was supposed to fix exactly this, and for the most part it does, but stock firmware still assumes a phone in every hand, a repeater on every hill, and an operator willing to babysit settings instead of walking.
FieldMesh is a firmware fork that starts from that field reality instead. Built on top of MeshCore by the fieldmesh-radio project, it turns four inexpensive LoRa boards into backpack nodes you can switch on and forget: position beacons that send themselves when you move, a one-press Off-Grid frequency where every node relays for the group, SOS alerts with a buzzer that works even when the phone is in a dry bag, messages you can read on the device screen, and a Station G2 box that becomes the camp base station with a message board, weather answers, and a WiFi dashboard for everyone’s phones. It is free, MIT licensed, and version 2.1.0 is a live, tested release based on MeshCore v1.18.0.
Last updated: November 2026. Current release v2.1.0, published 27 October 2026.
TL;DR: FieldMesh is a free MIT licensed MeshCore firmware fork for outdoor use, maintained by the fieldmesh-radio project and based on MeshCore v1.18.0. It adds battery-friendly GPS tracking with movement triggered adverts, a one-press Off-Grid mode on a separate regional frequency with Client Repeat so every node relays, two-step SOS alerts with a buzzer alarm that fires without a connected phone, on-device message reading plus message writing on the Wio Tracker L1, automatic identity backup, and a FieldMesh Repeater image that turns a Station G2 into a camp base station with a room server board, WiFi dashboard, and Open-Meteo weather answers. Four devices carry prebuilt firmware: the Elecrow ThinkNode M1, the Seeed Wio Tracker L1, the card sized T1000-E, and the Station G2. Install from the browser based FieldMesh Web Flasher in Chrome or Edge, back up your private key first if you come from a stock repeater or FieldMesh 1.x, and check local band rules before using Off-Grid outside the EU, UK, US, or Australia.
FieldMesh in its element: a tracker showing the SOS screen on the trail. Source: fieldmesh-radio/FieldMesh.
What FieldMesh Is
FieldMesh is a free, open source MeshCore firmware fork that adapts inexpensive LoRa tracker boards and the Station G2 for outdoor group use, adding battery-friendly GPS tracking, a one-press Off-Grid relay mode, SOS alerts, standalone on-device messaging, and a camp base station repeater image.
Unpack that sentence and the whole project comes into focus. The base layer is MeshCore, the LoRa mesh platform whose architecture splits the job two ways: a small companion radio speaks the mesh protocol over the air, and a phone app provides the keyboard, the map, and the contacts. Repeaters extend range, and room servers act as bulletin boards. FieldMesh keeps all of that compatibility and changes what the firmware does when the phone is away, the infrastructure is absent, or someone is in trouble. Your node sends its position when it matters instead of on a timer, your whole group can jump to a private frequency with one press where every radio relays for the others, an SOS goes out with your coordinates and battery level and wakes every receiver with a siren, and the camp box keeps the message board, the SOS list, and the weather forecast available to any phone on its WiFi, with or without internet in the field.
Three facts define the project. First, it is outdoor first in a literal sense: the feature list reads like a packing list for festivals, hiking groups, off-grid events, and tours, and the hero image on the repository is a hand holding a tracker showing an SOS screen above a mountain valley. Second, it is honest about scope: four tested devices get prebuilt firmware, everything else builds from source untested, Off-Grid knows EU, UK, US, and Australian frequencies only, and the README lists every limitation plainly. Third, it’s a healthy one maintainer project with open development: created in March 2026, moderately starred at 39 stars with 3 forks at the time of writing, MIT licensed, developed with Claude as a coding assistant with the collaboration visible in the commit history, planned from written specs, unit tested, CI built for all devices, and documented on a level most firmware twice its age never reaches.
Why a MeshCore Fork Exists
To understand why FieldMesh exists, start with what MeshCore already does well. MeshCore is an open LoRa mesh messaging system: small battery powered radios exchange short text messages across kilometres without any carrier network, using LoRa chirp spread spectrum in licence-exempt ISM bands. The design separates the radio from the interface. The companion radio handles the mesh protocol, flooding, contacts, and adverts, while the phone app handles typing, maps, and rooms. Fixed repeaters with better antennas and permanent power relay traffic across valleys and towns, and room servers give groups a shared channel with history. For neighbourhood meshes and always-on infrastructure, it’s an elegant split.
The field exposes the gaps in that split, and the FieldMesh author wrote them down as a feature list. A tracker that reports position on a fixed timer wastes battery and litters the mesh with packets nobody needed. A group that hikes out of repeater range has no one-press way to form its own relaying party net. An emergency alert that depends on a connected phone is no emergency alert at all. A node whose screen can’t show an incoming message is a beeper, not a communicator. And a base camp with a repeater still needs somewhere for the plan of the day, the weather, and the SOS list to live where every phone can see them. FieldMesh answers each of those with one focused mechanism: movement triggered GPS adverts, Off-Grid mode with Client Repeat, standalone SOS with a buzzer, on-device message history, and the FieldMesh Repeater.
The wider context matters too. The better known Meshtastic project covers similar ground with its own protocol, managed flood routing, MQTT uplinks, telemetry modules, and a large multi-vendor hardware ecosystem. MeshCore chose a different shape: chat first, with contacts, rooms, and repeaters as first class citizens and a phone app as the primary interface. FieldMesh doesn’t relitigate that choice. It takes the MeshCore shape as given and tunes it for the backpack, the festival wristband lane, and the convoy, where the phone may be off, the infrastructure may not exist, and the group still needs to find each other and call for help.
Supported Devices and Release Files
FieldMesh ships prebuilt firmware for four tested boards. The trackers pair with the standard MeshCore app exactly like stock companions, and every FieldMesh feature also works with no phone attached.
Device
What it is
Release file
Elecrow ThinkNode M1
Tracker with E-Ink display, GPS switch, one button, buzzer
ThinkNode_M1_companion_radio_ble_FieldMesh...zip
Seeed Wio Tracker L1
Tracker with OLED display and joystick, can write messages on-device
WioTrackerL1_companion_radio_ble_FieldMesh...zip
Seeed T1000-E
Card sized tracker with no display, one button, buzzer
t1000e_companion_radio_ble_FieldMesh...zip
Station G2
FieldMesh Repeater, the base station for a camp or tour
Station_G2_fieldmesh_repeater_FieldMesh..._full.bin
The Elecrow ThinkNode M1 is the most communicative of the three trackers thanks to its E-Ink screen, which stays readable in sunlight and sips power. The Wio Tracker L1 trades that for an OLED plus joystick, and the joystick is what makes on-device message writing possible. The T1000-E is the minimalist: no screen at all, card sized, one button, and a buzzer, built for a pocket or a bag strap where it just needs to beacon, listen for SOS, and beep when spoken to. The Station G2 is a different animal entirely, a mains or power station fed box that becomes the camp repeater, message board, SOS display, weather oracle, and WiFi dashboard host.
Other boards build from source but ship no prebuilt images and carry no hardware tested promise. The documented candidates are the Heltec Mesh Node T1, the SenseCAP MeshTracker X1, the Wio Tracker L1 E-Ink variant, and the Wio Tracker L1 1W, and single-button screenless devices in the spirit of the RAK WisMesh Tag could plausibly work like the T1000-E. If you prove one out, the project explicitly asks to hear about it in the Discussions hardware category.
Install and Update Without Losing Your Identity
Installation is deliberately boring, which is exactly what you want the night before a trip. The recommended path is the FieldMesh Web Flasher, a browser page that flashes the current release over USB. Open it in Chrome or Edge on a Windows, macOS, or Linux computer, because flashing needs WebUSB and Web Serial support that phones and Safari don’t provide. Connect the board with a real USB data cable, not a charge-only lead, and follow the two or three steps on screen. ThinkNode M1, Wio Tracker L1, and T1000-E take a Start update mode step followed by Flash, picking the port again since its name can change between steps. The Station G2 takes a straight Flash, and if the G2 isn’t found you hold its download button, tap reset briefly, release, and click Flash again. The flasher installs the current release for first installs and updates alike, and it never erases the device.
One warning deserves its own paragraph because it bites exactly the people with the most to lose. Back up your private key before a first install if you come from a stock MeshCore repeater on the Station G2 or from FieldMesh 1.x on a tracker. The FieldMesh Repeater uses a larger app partition, so moving from a stock repeater can lose the identity and settings; run get prv.key on the USB serial console and keep the output, then restore with set prv.key. Trackers arriving from FieldMesh 1.x, formerly numbered V5.x, should export the private key in the MeshCore app first. Updates inside the FieldMesh 2.x line keep settings, contacts, and identity, and exporting the key in the app before any first install is simply good habit regardless.
The manual path exists for operators who prefer it. Grab the file for your board from the latest release, flash tracker .zip files with the MeshCore Flasher as custom firmware, and write the Station G2 ..._full.bin at address 0x0 with esptool-js, never with erase. Repeater updates after that get easier: once the box runs, open the dashboard, go to Setup, log in, and upload the ..._update.bin under firmware update, with settings, room, team, and identity preserved and an automatic restart. USB reflashing with the full image always works as the fallback, which is why the repeater keeps both paths.
Since version 2.1 the trackers also carry their own safety net. FieldMesh keeps a copy of the node identity (/_main.id) and settings (/prefs.json) in a second filesystem, external QSPI flash on the ThinkNode M1 and Wio Tracker L1 and a separate area of internal flash on the T1000-E, restoring it automatically at the next boot if the main copy is lost. A factory reset from the app or CLI clears the backup too, which is how you genuinely start over. BLE pairings aren’t part of the backup, so expect to re-pair the phone.
Using Your Tracker in the Field
The on-device interface is the heart of the standalone promise, and it differs by hardware because the hardware differs. ThinkNode M1 and Wio Tracker L1 carry displays with the same page order, left to right: Messages, Recent, Tracking, Radio, GPS, Sensors where available, Repeater discovery, and SOS. The controls map to what each board physically has.
Control
ThinkNode M1, one button
Wio Tracker L1, joystick
Next / previous page
Short press / double press
Joystick right / left
Page action
Long press
Long joystick press, short press on Messages, GPS, and Repeater discovery
Buzzer on / off
Triple press
Triple press on the back button
Back
None
Back button
Each page action does something useful rather than decorative. Messages opens the message menu with all messages, favourites only, and Send Message on the L1. Recent sends an advert immediately. Tracking opens Outdoor Settings with GPS-Share and Off-Grid toggles. GPS toggles the receiver. Repeater discovery scans for repeaters in direct range. SOS opens the two-step send screen. In the message history a short press steps to the next message and a long press clears all; on an SOS alarm a long press acknowledges it; and a long press within the first 8 seconds after power-on enters CLI rescue mode inherited from upstream.
The ThinkNode M1 adds one hardware wrinkle worth memorising: its GPS runs only when the physical GPS switch and the GPS setting agree, and flipping the switch on also flips the setting on. That two-gate design is what makes multi-day battery life achievable, because the receiver is truly off rather than merely quiet.
The displayless T1000-E compresses everything into button press counts, and the manual is worth rehearsing before you need it for real.
Button
Action
1 press
Acknowledge an SOS alarm
2 presses
Send an advert now
3 presses
Buzzer on / off
4 presses
GPS on / off, saved across restart
5 presses
Off-Grid on / off, saved across restart
Long press
Power off, or CLI rescue mode within 8 s of power-on
GPS-Share tracking on the T1000-E is toggled from the MeshCore app advert settings, and note the asymmetry: the T1000-E receives SOS alarms but can’t send one, because sending needs the display’s two-step confirmation. Every action answers with a tone even when the buzzer is muted: ascending for on, descending for off, double beep for advert sent, and a continuous siren for an incoming SOS that you silence with one press. The Off-Grid state check is charmingly analogue: press five times and listen, ascending means it was off and is now on, descending means the reverse, then press five times again to return to where you started.
Battery-Friendly GPS Tracking
FieldMesh separates two things that stock MeshCore conflates: when you last heard from a node versus when that node last had a GPS fix. The RECENT page shows every recently heard node. The TRACKING page shows only nodes that actually sent a position, with the distance to you, green for fresh under five minutes and yellow for older. Only contacts starred as favourites in the MeshCore app appear there, capped at three, which keeps the page readable on a tiny screen and keeps the mesh quiet, since every tracked position costs airtime.
The GPS duty cycle borrows the approach Meshtastic users will recognise: while GPS is on, the receiver takes a fix every two minutes and sleeps between fixes, with search time adapting to visible sky and every constellation the chip supports in play across GPS, GLONASS, Galileo, and BeiDou. On the T1000-E that delivered about two to two and a half days with GPS on in the author’s tests, honestly flagged as having no exact current measurement. Boards whose GPS can’t be switched off simply keep it running.
Tracking itself, the GPS-Share switch, sends position only when it matters instead of on a fixed timer.
Situation
What is sent
First fix after GPS-Share or GPS switched on
Advert with position
Moved at least 100 m since last sent position
Advert with position, at most every 2 min
Moving fast, at least 30 km/h by car or bike
Fix every 60 s, adverts at most every 60 s
Nothing changed for 10 min
Heartbeat advert, with position only if the last fix is at most 2 min old
Two design choices keep the wider mesh healthy. In normal mode these adverts go out zero-hop, reaching only nodes in direct range and adding zero load to the wider network, while in Off-Grid mode they flood so the whole group sees them. And a position is only ever included when the fix is at most two minutes old, so the map never shows stale or cached coordinates as if they were live.
Off-Grid Mode: Your Group’s Private Frequency
Off-Grid mode is the headline feature and the one most worth understanding precisely. One press moves your node to a separate Off-Grid fre

https://hamradio.my/fieldmesh-meshcore-firmware-off-grid-gps-sos/

#gpstracking #meshcorefirmware #sos #fieldmesh #station #wiotracker #repeater #mit

#rss #feeds #linksharing #qwibbit

Secure Connected Devices: PC, Mac, Synology NAS, VPS and IoT ...

2026-11-04T05:10:50Z

Secure Connected Devices: PC, Mac, Synology NAS, VPS and IoT Guide Against Cyber Attacks

Your home today is a small data center. A Windows PC or Mac on the desk, a Synology NAS humming with family photos and backups, a five dollar VPS running a blog or a game server, a dozen always online gadgets from IP cameras to smart plugs to a robot vacuum. Every one of them is reachable, scannable, and attacked daily by automated botnets that never sleep. In the last three weeks alone, ransomware crews have encrypted hospital systems and NAS boxes, hackers have remotely heated supermarket freezers to 50C and started fires across two European attack waves, a PHP flaw put millions of servers in the crosshairs, and a hijacked army of routers and cameras fired the largest DDoS blast ever recorded. None of the victims were special targets. They were simply reachable and unpatched.
TL;DR: October and November 2026 brought a brutal reminder that basic hygiene stops most attacks: C3RB3R ransomware is encrypting outdated Synology NAS boxes through a flaw patched in December 2024, hackers remotely forced supermarket freezers up to 50C across France and then seven more European countries, the PHP2Shell flaws CVE-2026-109618 and CVE-2026-109619 allow unauthenticated remote code execution on millions of PHP servers, the Byod gang forced a New Mexico hospital to divert ambulances, and the Aisuru IoT botnet fired a record 22.2 Tbps DDoS blast. Protect yourself by updating everything and enabling automatic updates, blocking NAS and camera management pages from the internet, using unique passwords with a manager plus two-factor authentication everywhere, keeping 3-2-1 backups with one offline copy and testing restores, hardening SSH and PHP on any VPS, and isolating IoT gadgets on a separate network with default credentials changed. The steps below walk through Windows, Mac, Synology, VPS, and smart home gear one by one.
Synology DiskStation Manager. Outdated DSM is exactly what the November 2026 C3RB3R ransomware wave exploits. Screenshot: synology.com.
What Device Hardening Is
Device hardening is the practice of reducing every computer, phone, NAS, server, and smart gadget you own to its smallest attackable surface by updating software, closing unneeded network access, enforcing strong unique authentication, encrypting data, and keeping tested backups. So that automated mass attacks simply move on to easier targets.
That definition carries the whole strategy, so unpack it slowly. Attackers today rarely pick you personally. Botnets like Mirai, Aisuru, and Kimwolf sweep the entire internet trying factory passwords such as admin/admin, known flaws in old firmware, and open management ports, exactly the way the Trend Micro research describes hundreds of thousands of routers and cameras conscripted into record DDoS attacks. Ransomware crews buy access from initial access brokers who got in through one unpatched VPN box or one reused password, as the October 2026 hospital and university attacks keep demonstrating. Hardening doesn’t mean buying security products. It means making your devices boring to attackers: patched, closed, authenticated, encrypted, and backed up. Each section below applies that definition to one category of gear, with the highest value actions first.
The Threat Landscape Right Now: October and November 2026
Security advice lands harder when you can see the ambulances. Here is what actually happened in the four weeks before this article, and what each event teaches a home or small business owner.
C3RB3R ransomware encrypts Synology NAS boxes, November 2026
Taiwanese vendor Synology warned customers in early November 2026 that C3RB3R ransomware, also called Cerber or CerberImposter, is actively encrypting NAS devices running outdated DiskStation Manager software, as reported by BleepingComputer and The Hacker News. Victims find a ransom note named read-me3.txt and every file renamed with a .L0CK3D extension, with demands pointing to a dark web portal selling a C3rb3r Decryptor for thousands of dollars. The attackers exploit a vulnerability patched back in December 2024, linked to Synology advisory SA 24:20 covering CVE-2024-10441 with a CVSS score of 9.8 plus file read and file write flaws. Hong Kong’s CERT issued its own ransomware alert telling owners to isolate infected boxes and preserve logs. Threat watchdog Shadowserver tracks more than 380,000 internet exposed Synology devices. The lesson is the oldest one in security and still the most ignored: the patch existed for almost a year, and the victims are the ones who never installed it. This section’s NAS checklist below is written directly against this attack.
PHP2Shell puts millions of servers in the crosshairs, October 2026
On October 21, 2026, the PHP project shipped emergency patches for two critical remote code execution flaws, CVE-2026-109618 and CVE-2026-109619, collectively called PHP2Shell, both scoring 9.2 out of 10, as covered by BleepingComputer. Either flaw lets an unauthenticated attacker run code on the server. Fixed releases are PHP 8.2.35, 8.3.36, 8.4.27, and 8.5.12. Amazon confirmed exploitation began four days before patches existed, Cloudflare recorded millions of attempts, and the US CISA added both flaws to its Known Exploited Vulnerabilities catalog on October 26, 2026, per BleepingComputer’s follow-up. Singapore’s Cyber Security Agency published alert AL-2026-151 urging immediate patching of internet facing systems. A related flaw, CVE-2026-115560, still has no patch at all. If you run WordPress, Nextcloud, or any PHP app on a VPS, this is your wake-up call, and the server section below tells you exactly what to update. Full disclosure: this very site, hamradio.my, runs on PHP 8.4.27, the fixed release, because server software is only as safe as its last update.
Hospitals, railways, and universities under ransomware fire
On October 26, 2026, the University of New Mexico Health Sciences Center disclosed a cyberattack that forced ambulance diversions, suspended surgeries, and cancelled classes, with a new gang called Byod claiming it hit 36,000 systems, according to BleepingComputer. The same month brought ransomware to Japan’s Keio railway, the IDCF cloud serving government clients, and Osaka Metropolitan University, plus 200,000 people potentially exposed at Denmark’s Technical University, as summarized in CM Alliance’s October 2026 roundup. Developer platform Zeabur announced it would shut down entirely after attackers used leaked plaintext credentials, a Russian hoster leaked customer data, hackers remotely manipulated French supermarket refrigeration hard enough to start fires, and an incident at BAE Systems disrupted US federal court operations. Attackers don’t care whether you are a hospital or a hobbyist. They care whether the door is open.
IoT botnets fire a 22.2 Tbps record blast
In late October 2026, an IoT botnet hammered a European network provider with a 22.2 Tbps UDP carpet bombing attack from more than 404,000 compromised devices across 14 networks, over twice the previous 11.5 Tbps record, as reported by CircleID. Investigators point at the Aisuru botnet, built from routers, DVRs, and embedded devices hijacked through known flaws. Backbone operator Arelion found Aisuru drives about one third of all DDoS traffic on its network, with more than 500,000 enslaved IoT and Android TV devices. Sibling botnet Kimwolf infected over 2 million Android TV boxes. In March 2026 the FBI and partners in Canada and Germany seized infrastructure behind four such botnets controlling over 3 million devices. Barracuda’s botnet briefing traces the fuel to the same causes every year: default credentials, unpatched firmware, and exposed Telnet, HTTP, and ADB debug interfaces. Your camera or router is either patched behind your firewall, or it is somebody else’s cannon. The IoT section below makes sure it’s the former.
When Hackers Turn Freezers Into Ovens: the 2026 Supermarket Fridge Attacks
The supermarket fridge attacks are the clearest proof this year that an internet connected device with old software and remote management can cause physical destruction, with hackers remotely heating store freezers to 50C, starting fires, and forcing millions of euros of food to be destroyed across Europe in two attack waves.
Wave one hit France on the night of Sunday October 11, 2026. Attackers hijacked refrigeration controllers running older Danfoss control software and triggered the built-in automatic defrost cycle, driving fridges and freezers up to 50C for around six hours overnight, according to BleepingComputer. Managers arriving on Monday morning walked into waves of hot air. Some units caught fire. Around 150 to 300 Intermarche and Netto stores plus almost 100 Cooperative U outlets were hit, alongside Carrefour, Monoprix, Leclerc, and Super U locations, with 250 to 350 stores affected in total. Every tray of fresh and frozen food had to be thrown away. Two supermarkets in Alsace alone reported combined losses of 270,000 euros, putting the nationwide damage in the millions. The Paris prosecutor opened a large scale cyberattack investigation, and Al Jazeera quoted retailers calling it unprecedented.
Wave two landed on November 2, 2026, three weeks later, and proved the first attack was a rehearsal. Dozens of supermarkets across at least seven European countries reported the same overnight warming, per the BBC and Euronews: around 40 Delhaize stores in Belgium and Luxembourg, ICA and Coop shops in Sweden with freezers set to 50 degrees, K-stores in Finland, plus outlets in Ireland, the Netherlands, Estonia, Italy, and Portugal. The method matched, evaporators forced into defrost mode through remote maintenance access. Belgian refrigeration operator Sabcobel admitted it had been racing to secure units since the French attack but that some needed physical modifications, not just software patches. EU Agriculture and Food Commissioner Christophe Hansen called it a deliberate attempt to disrupt the food supply chain. Investigators are even examining whether freezer failures at American military bases in August connect to the same campaign.
Three lessons for your own home fall straight out of this story. First, remote maintenance interfaces are the target. Supermarket fridges are online so technicians can service them without driving out, exactly like the cloud viewing on your IP camera or the vendor app on your smart thermostat. Every such interface needs unique credentials, current firmware, and ideally no direct internet exposure at all. Second, old controller software is the entry point. The French controllers ran outdated Danfoss versions, the same pattern as the outdated DSM behind the Synology ransomware wave and the unpatched PHP behind PHP2Shell. If the vendor stops shipping fixes, replace the device. Third, segmentation limits the blast radius. Stores that could isolate refrigeration controls from the compromised management service stayed cold. Your IoT network separation below does the same job at home: a hijacked camera should never be able to reach anything that matters.
Secure Your Windows PC
Windows remains the most attacked desktop because it is the most common one, and most compromises arrive through the same few doors: phishing mail, malicious downloads, unpatched software, and reused passwords. Close those doors in this order.
Update Windows and apps automatically. Leave Windows Update on automatic, including driver and Microsoft Store updates, and enable automatic updates in browsers, Office, and third party tools. Attackers weaponize patched flaws within days, as the PHP2Shell timeline proved at server scale and monthly Patch Tuesdays prove on desktops. Reboot when asked. A pending update protects nothing.
Leave Microsoft Defender on and SmartScreen enabled. The built-in antivirus, firewall, and SmartScreen download reputation checks are genuinely good now. Don’t disable them for a “lighter” free antivirus from a pop-up ad, which is itself a classic infection vector. Run a full scan monthly.
Use a standard account daily, keep one separate admin. Most malware inherits your privileges. Create a standard user for everyday work and keep the administrator account only for installs. This single change neuters a large share of infostealers.
Encrypt the drive with BitLocker. On Pro editions turn on BitLocker, on Home use device encryption where offered, and save the recovery key to your Microsoft account and on paper. A stolen laptop with an unencrypted disk is a data breach with no hacking required. More detail is on the Windows security pages.
Lock down remote access and sharing. Turn off Remote Desktop unless you actively use it, disable network discovery and file sharing on public networks, and treat every unexpected “invoice”, “parcel”, or “account locked” attachment as hostile. Verify the sender through a second channel before clicking.
Back up to two places, one offline. Keep File History or a system image on an external drive that you disconnect afterwards, plus an offsite copy. Ransomware can’t encrypt a disk sitting in a drawer. Test a file restore twice a year.
Secure Your Mac
Macs benefit from strong defaults, Gatekeeper app signing checks, XProtect malware signatures, and rapid security responses, but “Macs don’t get malware” has been false for a decade. Infostealers, fake updates, and trojanized developer tools target Mac users precisely because they feel safe. Apple documents the platform protections in its macOS pages.
Install updates and Rapid Security Responses immediately. System Settings, General, Software Update, with automatic updates fully enabled. These ship between major releases and close actively exploited WebKit and kernel flaws.
Turn on FileVault and the firewall. FileVault encrypts the whole disk, which protects a lost MacBook completely. Enable the built-in firewall with stealth mode so the Mac ignores unsolicited probes on hostile networks.
Keep Gatekeeper and SIP at their defaults. Install software from the App Store or signed developer disk images. Never disable Gatekeeper or System Integrity Protection to run a “cracked” app. That single bypass is how most Mac malware gets its foothold, and trojanized tools, like the recent poisoned AI coding extension that reached thousands of developers, thrive on users who override warnings.
Use a standard account, Find My, and Safari’s fraud warnings. Daily driving as admin hands every malicious script the keys. Find My plus Activation Lock makes a stolen Mac unsellable and unusable. Leave Safari fraudulent website warnings and iCloud Private Relay options on.
Back up with Time Machine to a disk you rotate, plus offsite. A USB drive that never leaves the desk dies in the same fire, flood, or burglary as the Mac. Rotate two drives or add an encrypted cloud copy, and verify a restore by actually opening an old file.
Consider Lockdown Mode only if you need it. Journalists, activists, and high risk targets should enable Lockdown Mode, which hardens messages, browsing, and attachments at the cost of convenience. For everyone else, the defaults plus the steps above are the right balance.
Secure Your Synology NAS
Read this section twice, because a NAS is the worst thing in your house to lose: it holds the originals and, too often, the only backups. The November 2026 C3RB3R wave exists for exactly one reason, outdated DSM reachable from the internet. Synology’s own security advisories and the DSM update pages are your standing references.
Update DSM today and enable automatic updates. Control Panel, Update and Restore, with automatic installation of important updates. C3RB3R exploits a flaw fixed in December 2024, so anyone current was never at risk. Check quarterly that auto update actually ran.
Take the management page off the internet. This is the highest value NAS action there is. Remove port forwards for ports 5000 and 5001 on your router, disable UPnP rules that expose them, restrict the DSM login to your LAN in Control Panel Security Firewall, and avoid QuickConnect exposure unless you need it. Administer remotely over a VPN such as Tailscale or WireGuard instead. More than 380,000 Synology boxes sit internet exposed right now, and scanners find them in minutes.
Kill the default admin and enforce 2FA. Disable the built-in admin account, create a separate administrator wi

https://hamradio.my/secure-connected-devices-pc-mac-nas-vps-iot-guide/

#synologynas #devicehardening #vps #mac #attacks #php #iot #ransomware #windows #november

#rss #feeds #linksharing #qwibbit

Serial Studio: Open Source Telemetry Dashboard for Arduino, ...

2026-11-03T19:16:19Z

Serial Studio: Open Source Telemetry Dashboard for Arduino, ESP32, and Ham Benches

TL;DR: Serial Studio telemetry dashboard turns raw bytes from Arduino, ESP32, STM32, Raspberry Pi, and industrial buses into a live dashboard with plots, gauges, GPS maps, FFT, and 3D views, plus CSV logging and command-back controls. The GPL-3.0 core covers UART, Bluetooth LE, TCP, UDP, WebSocket, and HTTP with 17 widgets free forever on Windows, macOS, Linux, and Raspberry Pi. Pro adds MQTT, Modbus, CAN bus with DBC import, OPC UA, Siemens S7, EtherNet/IP, IEC 60870-5-104, audio, USB, HID, MDF4, SQLite session recording, PDF reports, and an AI project assistant. For hams it replaces four separate tools: serial monitor, serial plotter, GPS tracker map, and CSV logger, in one binary you can run on the bench PC and the field laptop.
What Serial Studio Is
Serial Studio is a free open source telemetry dashboard that connects to a microcontroller, instrument, or industrial bus, decodes one user-defined data format, and renders live plots, gauges, maps, and 3D views around it while recording every frame for replay and export.
The project lives at https://github.com/Serial-Studio/Serial-Studio and ships official binaries from https://serial-studio.com/. It is developed and maintained by Alex Spataru, holds about 7,300 stars and 1,100 forks, and has more than 3,800 commits on master. The product site lists version 4.2.1 as current, with a 14-day Pro trial that starts the first time a Pro feature is used and falls back to the free tier instead of locking the user out.
Serial Studio live dashboard with plots, gauges, and project editor. Screenshot: Serial Studio repository, GPL-3.0.
Project Snapshot
Item
Detail
Project
Serial Studio by Alex Spataru
License
Dual license: GPL-3.0-or-later core, commercial Pro modules and official binaries
Latest release
4.2.1 (sealed deployment packages, ISA-18.1 alarm annunciator, trial on demand)
Stars and forks
About 7,300 stars, 1,100 forks, 133 watchers
Platforms
Windows 10 and 11, macOS 13 plus Universal, Linux x64 and ARM64, Raspberry Pi 4 and 5 64-bit
Engine
Native C++ on Qt 6.9 plus, 60 fps, parse pipeline benchmarked at 256,000 frames per second as a CI gate
Free drivers
Serial and UART, Bluetooth LE, TCP, UDP, WebSocket, HTTP
Pro drivers
MQTT with Sparkplug B, Modbus TCP and RTU, CAN bus, OPC UA, Siemens S7comm, EtherNet/IP, IEC 60870-5-104, audio input, raw USB libusb, HID hidapi, Process I/O
Free widgets
17 including line plot, multi-plot, gauge, bar, compass, meter, GPS map, FFT, accelerometer, gyroscope, data grid, LED panel, terminal, web view, plus Clock and Stopwatch
Pro widgets
3D Plot, Waterfall spectrogram, Image View live camera, Canvas scripted widget, output buttons, sliders, knobs, Output Panel, Notification Log
Scripting
Built-In form parsers, JavaScript, LuaJIT 2.1 with 5.1 syntax, 28 protocol templates
Recording
CSV in free build, MDF4 read and write, SQLite Historian, HTML and PDF session reports in Pro
Automation
TCP API with 360 plus commands on port 7777, gRPC server on port 8888, MCP server for Claude Desktop
Links
GitHub repo, product site, help center, releases, pricing
Why the Arduino Serial Plotter Is Not Enough
Every Arduino and ESP32 builder knows the default loop. Open the Arduino IDE Serial Monitor to see numbers scroll. Open the Serial Plotter to see one wobbly line. Open a terminal like PuTTY or minicom to log to a file. Open Excel or Python later to make a chart anyone would believe. Four tools, four windows, zero memory of what the test meant a week later.
That stack breaks down the moment the project grows past one sensor. The IDE plotter draws every column the same color with no units, no gauges, and no map. PuTTY logs bytes but understands nothing inside them. MATLAB, LabVIEW, and Grafana understand everything but demand licenses, servers, and setup days. Industrial SCADA understands PLCs but has no interest in a ten-dollar ESP32 on a breadboard.
Serial Studio sits in the gap. One project file describes the frame format once. The same binary draws the dashboard, logs synchronized CSV, replays the session later through the same widgets, and sends commands back over the same cable. The free build already covers the whole hobby bench. The Pro modules extend the same pipeline to the factory floor without changing how the app feels.
How It Works: Six Stages, One Pipeline
Every device flows through the same six stages, documented in the official data flow reference. Learning this order makes every later feature obvious.
1. Connect
The driver receives raw bytes. In the free build that means UART serial, Bluetooth LE GATT notifications, or a network socket over TCP, UDP, WebSocket, or HTTP. Pro adds the industrial and media transports listed in the snapshot table.
2. Detect frames
The Frame Reader finds each frame in the byte stream. Optional start and end sequences plus CRC validation catch corrupted data before it ever reaches the screen, which matters on noisy RF links and long RS-485 runs.
3. Parse frames
Quick Plot mode splits comma, semicolon, tab, space, or name-colon-value lines automatically with no project at all. Project mode runs a parser: a Built-In template configured through a form, or a JavaScript or Lua parse() function for binary protocols, checksums, and packed bitfields.
4. Populate widgets
Dataset values flow into widgets. A plot shows history, a gauge shows now, a GPS map shows where, a data grid shows exact numbers, an LED panel shows status bits. One dataset can feed several widgets at once.
5. Record
In parallel, every frame is written to disk. CSV in the free build. SQLite session database and MDF4 in Pro. Recording never slows the live view because it runs on its own path.
6. Replay and export
Open any recording back through the same dashboard. Scrub the timeline, compare runs, and export a styled HTML or PDF session report with interactive charts when the test is ready to share or sign.
Connect to Anything: The 14 Drivers
The driver list is the clearest free versus Pro boundary, so read it before planning a project.
Driver
Edition
What it talks to
UART and Serial
Free, GPL
USB-UART, RS-232, RS-485, TTL from Arduino, ESP32, STM32, Teensy, Raspberry Pi
Bluetooth LE
Free, GPL
GATT peripherals, sensor tags, wireless loggers
TCP and UDP
Free, GPL
Unicast, broadcast, multicast streams from any host
WebSocket and HTTP
Free, GPL
Web dashboards, REST pollers, simple bridges
MQTT with Sparkplug B
Pro
Any broker, TLS and auth, host or edge-node roles
Modbus TCP and RTU
Pro
Holding, input, coil, discrete registers, vendor CSV, XML, JSON map import
CAN bus
Pro
Classic CAN and CAN FD, DBC import, J1939 and ISO-TP reassembly
OPC UA
Pro
Browse address space, tick tags, signed and encrypted sessions
Siemens S7comm
Pro
S7-300, S7-400, S7-1200, S7-1500 by absolute address over ISO-on-TCP, read-only
EtherNet/IP
Pro
Allen-Bradley ControlLogix, CompactLogix, MicroLogix, Micro800, SLC 500, PLC-5 plus Omron NJ and NX by tag name, read-only
IEC 60870-5-104
Pro
RTUs and substation gateways, general interrogation then spontaneous reports, read-only
Audio input
Pro
Any sound card as multichannel oscilloscope, FFT, or waterfall
Raw USB and HID
Pro
libusb bulk and isochronous endpoints, hidapi gamepads and custom devices
Process I/O
Pro
Spawn a script and pipe stdout, or read a named pipe from Python or Rust
Three details matter for bench users. First, the three PLC drivers are read-only clients, so connecting Serial Studio to a running machine can’t change its state. Second, several Pro drivers ship with simulators, so OPC UA, Modbus, S7, IEC 104, and Sparkplug can be evaluated with no hardware on the desk. Third, multi-device projects, where each source keeps its own protocol, are a Pro feature, while one device at a time is free without limit.
Visualize Everything: 20 Plus Widgets
The free build ships 17 widgets, and Pro adds the rest past 20. Four of them deserve special attention because they behave differently from what newcomers expect.
Bar, Gauge, Compass, and Meter each render as a two-page swipe view. Page zero is the analog face for glanceable trend. Page one is a large digital readout for the exact value. One tile therefore covers both the across-the-room check and the write-it-in-the-logbook number.
The headline widgets group naturally:
Time and spectrum: line plot, multi-channel plot, FFT spectrum, Waterfall spectrogram in Pro, XY scope patterns through the Canvas widget in Pro.
State at a glance: gauge, bar, meter, compass, LED panel, data grid, bar panel.
Motion and position: GPS map with OpenStreetMap tiles, accelerometer, gyroscope, 3D Plot in Pro.
Text and web: terminal console with VT-100 and ANSI color, Quick Plot console, web view for embedding pages, Clock and Stopwatch utilities.
Media and custom: Image View live JPEG or PNG camera stream in Pro, Canvas widget in Pro driven by a JavaScript paint(ctx, w, h) callback with 18 templates including oscilloscope, polar plot, artificial horizon, audio VU, dial gauge, heatmap, vector field, and sparklines.
Command back: buttons, toggles, sliders, knobs, text fields, and freeform Output Panel in Pro, with templates that emit GCode, SCPI, Modbus, NMEA, and CAN frames on demand or on a timer.
Custom workspaces, also Pro, split large projects into focused tabs with a searchable taskbar, which is how a 200-dataset factory project stays navigable.
Parse and Transform Without Losing the Raw Data
Parsing is where Serial Studio earns its keep over a dumb plotter, and it offers three levels.
Built-In templates are compiled C++ parsers configured through a form. Delimited text, key-value pairs, JSON, NMEA, Modbus, and TLV shapes need no code at all, and this is the default for new projects.
JavaScript and Lua parsers handle everything else. A short parse() function receives the raw frame bytes and returns datasets, so MAVLink, UBX, SiRF, RTCM, MessagePack, COBS, SLIP, YAML, INI, and custom binary checksums all decode the same way. Twenty-eight script templates ship in the box as starting points, and existing protocol code can usually be pasted in with small edits. Lua runs on LuaJIT 2.1 with 5.1 syntax and compatibility shims.
Per-dataset transforms, a Pro feature, run every frame as short JS or Lua snippets for EMA smoothing, scaling, calibration coefficients, and unit conversion. Variables act as a shared bus so transforms can derive computed datasets from each other, and Data Tables hold constants and system values in one place. Every session keeps both the raw and the computed values, which is the traceability auditors and lab notebooks demand. Restore the project months later and the readings still match.
Record, Replay, Report, Transfer
Logging is split by edition, and the split is honest about who needs what.
CSV export in the free build writes every frame, ready for Excel, Python, MATLAB, or R. That alone replaces most bench logging.
Pro adds the formats professionals get asked for. MDF4 import and export covers CAN bus, LIN, FlexRay, and analog with files that open in CANalyzer, MATLAB, or Python. The Historian records frames and raw bytes into SQLite, then the Database Explorer browses, tags, exports, and replays full sessions. Session reports export a styled HTML or PDF with interactive Chart.js plots, which is the shortest path from a test run to a signed document. File transfer over the active connection speaks plain text, raw binary, XMODEM, YMODEM, and ZMODEM with CRC and resume, so firmware images and config files go to the device from the same console. A live InfluxDB 2.x sink streams every frame as line protocol for teams that already run a time-series stack.
Automate It: TCP API, gRPC, MCP, AI Assistant
Three machine interfaces expose the same control surface, plus one human-facing AI panel.
The TCP API on port 7777 offers more than 360 JSON commands for programmatic control, with a working API client example in the repo. The gRPC server on port 8888 mirrors the same command set over protobuf with live frame and raw-data streaming, documented in the official gRPC reference. An MCP server wraps the full surface for the Model Context Protocol, so Claude Desktop or any MCP host can drive Serial Studio, with a client example included.
The AI Assistant panel, a Pro feature, is a bring-your-own-key chat that edits the project file: sources, datasets, parsers, transforms, output widgets, and workspaces. Eight providers are supported: Anthropic, OpenAI, Google Gemini, DeepSeek, Groq, Mistral, OpenRouter, and local OpenAI-compatible endpoints such as Ollama, llama.cpp, LM Studio, and vLLM for fully offline use. Mutating actions show an Approve or Deny card first, and connection control plus device writes stay blocked until the Allow device control toggle is enabled as an Always-confirm action. Operator deployments hide the panel entirely.
Quick Start: Five Minutes from Zero to Plots
The official getting started guide covers the same path. This is the short version that works on any OS.
Download the build for the bench PC from the latest release. Windows runs an MSI installer, macOS drags a Universal DMG into Applications, Linux makes the AppImage executable and runs it, Raspberry Pi uses the ARM64 AppImage on a 64-bit OS.
Flash a two-sensor sketch to an Arduino Uno or any ESP32. The classic example prints two comma-separated values per line at 10 Hz.
Open Serial Studio, pick the serial port and baud rate, and click Connect.
Enable Quick Plot. Two traces appear with no project file at all.
When ready, open the Project Editor, define one group with two datasets, attach a line plot plus a gauge and a data grid, save the project, and reconnect through it.
void setup() {
Serial.begin(115200);
}
void loop() {
int temperature = analogRead(A0);
int humidity = analogRead(A1);
Serial.print(temperature);
Serial.print(",");
Serial.println(humidity);
delay(100);
}
Upload, select 115200 baud in Serial Studio, enable Quick Plot, and both channels draw immediately. Change the delay to 20 ms and watch the throughput headroom: the parse pipeline is CI-gated at 256,000 frames per second, so a 50 Hz bench stream is effectively idle.
Linux users have two extra options. The AppImage runs anywhere from the Debian 10 and RHEL 8 generation onward with its own bundled glibc, and Flathub carries a regularly updated Flatpak that tends to behave better on ARM64. On Wayland desktops expect the occasional missing window shadow, a known cosmetic quirk.
Serial Studio Compared with the Usual Tools
Tool
Cost
Inputs
Dashboard
Logging
Command back
Best for
Serial Studio free
Free, GPL-3.0
UART, BLE, TCP, UDP, WebSocket, HTTP
17 widgets, maps, FFT
CSV, replay
No
Bench telemetry with zero setup
Serial Studio Pro
Paid license, 14-day trial
Plus MQTT, Modbus, CAN, OPC UA, S7, EtherNet/IP, IEC 104, audio, USB, HID
Plus 3D, waterfall, camera, Canvas, outputs
Plus MDF4, SQLite Historian, PDF reports
Yes, buttons and sliders
Lab, automotive, factory floor
Arduino IDE Serial Plotter
Free
USB serial only
Single multi-line plot
None
No
First-blink sanity check
PuTTY, Tera Term, minicom
Free
Serial, SSH, Telnet
None, text only
Raw text capture
Manual typing
Console access and AT commands
Processing sketches
Free, code
Serial, network
Hand-built 2D
Hand-built
Hand-built
Custom one-off visualizers
MATLAB and LabVIEW
Licensed, costly
Broad with toolboxes
Full engineering plots
Full test management
Full
Funded labs and production test
Grafana plus broker
Free, server
MQTT and databases
Web dashboards
Time-series DB
Via plugins
Fleet dashboards, not bench debug
The pattern is consistent. Serial Studio wins when one engineer needs a real dashboard this afternoon with no server, no license ticket, and no custom GUI code. It loses when the job is fleet monitoring across months, where Grafana and a database are the right answer, or when certification demands LabVIEW test management.
How Amateur Radio Operators Can Use It
Serial Studio isn’t a ham radio program, but it speaks every protocol a ham bench produces. Each pattern below maps to real operating, building, or training.
1. ESP32 shack sensor node in an afternoon
Point an ESP32 at the shack: DHT22 temperature and humidity, BMP280 pressure, a current sensor on the 13.8 V supply, and an SWR bridge voltage. Print one CSV line per second over USB serial. Serial Studio draws temperature and voltage plots, a humidity gauge, and a data grid with exa

https://hamradio.my/serial-studio-open-source-telemetry-dashboard-arduino-esp32/

#serialstudio #opensource #telemetrydashboard #pro #free #alexspataru #raspberry #gps #csv #tcp

#rss #feeds #linksharing #qwibbit

When Soldiers Shop Civilian: 12 Civilian Products Modern ...

2026-11-03T17:55:36Z

When Soldiers Shop Civilian: 12 Civilian Products Modern Militaries Use Every Day

TL;DR: Modern armies run on civilian gear. Toyota Hilux pickups won the Toyota War, DJI camera drones watch over Ukraine, Starlink carries frontline comms, and smartphones, Xbox controllers, iPads and GoPros fill roles once held by custom military hardware. Twelve products inside, plus why generals buy off the shelf and what hams and preppers should copy.
What Civilian to Military Adoption Means
Civilian to military adoption means an armed force takes a product designed for the high street and puts it into uniform with little or no modification, a practice procurement officers call COTS, short for commercial off the shelf.
Everyone knows the traffic runs the other way. GPS, the internet, the microwave oven and duct tape all started in uniform and ended up in civilian kitchens and cars. This article walks the road in the opposite direction. The modern battlefield runs on pickup trucks built for farmers, camera drones built for weddings, game controllers built for teenagers, tablets built for airline passengers, watches built for construction workers, and internet satellites built for rural subscribers. In Ukraine, in the Red Sea, in the Pacific and at every NATO training ground, the most important equipment is often the equipment any civilian can buy on a credit card.
Why Generals Buy Civilian
Four pressures push every modern military toward the civilian shelf.
Speed. A purpose built military program takes 10 to 20 years from requirement to fielding. A civilian product ships next week. When quadcopter drones started deciding battles in Ukraine in 2022, neither side waited for a ten year drone program. They bought DJI Mavics by the thousand and flew them the same month.
Cost. A custom military handheld radio can cost 10,000 dollars per unit. A ruggedized Samsung phone running ATAK tactical software costs under 3,000 dollars with its software load and does mapping, messaging, drone feeds and blue force tracking in one device. When budgets tighten, that arithmetic wins every argument.
The innovation gap. In the Cold War, the military led in computing, materials and communications. Today the lead has flipped. Apple, Samsung, SpaceX, DJI and Garmin spend more on research in a year than most defence ministries. The best camera, the best screen, the best battery and the best satellite constellation are civilian products first.
Logistics. A Toyota Hilux can be repaired in any town on earth with parts from any market. A bespoke military truck needs a bespoke supply chain. Reliability plus global spare parts beats peak performance in every long war, which is exactly the lesson of the Toyota War of 1987.
The trade offs are real and this article won't hide them. Civilian gear can phone home to servers in rival countries, it can fail in dust and shock, and its supply chain can be cut in a crisis. Every product below earned its place, and every one carries a warning label. That tension is the story.
A civilian Toyota Hilux pickup of the type converted into armed technicals across Africa and the Middle East. Photo: Wikimedia Commons.
1. The Toyota Hilux Pickup and the Technical
If one civilian product symbolizes modern irregular war, it's the Toyota pickup.
The story peaks in the Toyota War, the final phase of the Chadian Libyan conflict in 1987. Chad, backed by France and the United States, mounted anti tank missiles and heavy machine guns on around 400 Toyota Hilux and Land Cruiser pickups and used desert mobility to destroy Libyan armor. Libya lost roughly one tenth of its army, about 7,500 troops, plus hundreds of tanks and vehicles. The war was named after the truck. Time magazine and the US Army war college still teach it as proof that mobility plus cheap firepower beats heavy armor in open terrain.
Why the Hilux kept winning for the next forty years comes down to engineering facts:
Ladder frame chassis with live axles that tolerate overloading and bad welding jobs in field workshops
Naturally aspirated and later turbo diesel engines that run on poor fuel and survive missed services
Leaf spring rear suspension that carries a ton of fighters, weapons and water without complaint
Global parts commonality, since the same truck sold from Bangkok to Nairobi to Lima
A steel bed that accepts a gun ring, a mortar baseplate or a stretcher with hand tools
Regular armies formalized what guerrillas discovered. The US Army Special Forces ran the S9 pickups, and in 2020 the Army selected the GM Defense Infantry Squad Vehicle, a 5,000 pound truck based 90 percent on the civilian Chevrolet Colorado ZR2 pickup, including its 2.8 liter Duramax diesel, its frame and many suspension parts. Nine soldiers, a truck that fits inside a CH-47 Chinook, built from showroom parts. The British Army did the same logic earlier with Land Rover, itself a 1948 civilian farm vehicle that became a military icon for fifty years.
Civilian lesson: the prepper affection for old diesel pickups isn't nostalgia. It's the same calculation the Pentagon made. Simple diesel, manual linkages where possible, parts shared with a million other trucks on the road, and a bed that works as a table, a bed and a gun platform.
2. DJI Camera Drones Over Every Trench
The DJI Phantom first flew in 2013 as a wedding videographer tool. A decade later its descendants owned the sky over Ukraine.
Both Ukrainian and Russian units fly thousands of off the shelf DJI Mavic 3 and similar quadcopters for artillery spotting, trench reconnaissance and bomb dropping with 3D printed release hooks. A 2,000 dollar drone with a 4/3 sensor and 46 minutes of endurance replaced missions that once needed a manned helicopter or a million dollar military UAV. First person view racing drones, another civilian hobby product costing a few hundred dollars, became precision strike weapons with wire guided immunity to jamming through fiber optic spools.
Militaries learned three hard lessons civilians should note:
Consumability beats capability. Losing a 2,000 dollar drone to find a 2 million dollar air defence system is a trade every commander takes. Units plan for losses per week the way they plan for fuel.
The software chain matters more than the airframe. DJI AeroScope remote ID broadcasts exposed operator positions, pushing Ukraine toward Autel, Parrot Anafi USA and home built FPV designs. Buying the drone is step one. Owning its data path is step ten.
Hobby skills transfer directly. FPV racing pilots, RC modelers and photography drone operators needed almost no retraining. The civilian hobby base became a trained reserve overnight.
Civilian lesson: for search and rescue teams, farmers and hams doing antenna site surveys, the same Mavic class drone is a legitimate capability. Learn the remote ID rules in your country, practice manual flight without GPS assist, and assume anything you fly near a conflict or an airport is tracked.
A DJI Phantom class camera drone in flight. Thousands of similar civilian quadcopters serve as reconnaissance and strike platforms in Ukraine. Photo: Wikimedia Commons.
3. Starlink Satellite Internet at the Front
Starlink was designed to sell broadband to rural homes, RVs and ships. It became the backbone of Ukrainian command and control within weeks of the 2022 invasion, with tens of thousands of terminals rushed in as fiber lines were cut and cell towers destroyed.
The military appeal is structural. A flat high performance dish, about the size of a pizza box, delivers 50 to 200 Mbps with 25 to 60 ms latency from low earth orbit, sets up in minutes, runs on a generator or a truck battery through an inverter, and moves with the unit. Artillery teams, drone pilots and headquarters share video, fire missions and intelligence over the same link a farmhouse uses for streaming. The US Department of Defense formalized the relationship with the Starshield service line for government users.
The cautions are equally structural. Civilian constellations are single points of failure owned by one company, subject to business decisions, export controls and cyberattack. Russia has targeted Starlink with jamming and attempted hacking, and service over Taiwan, mainland China and other regions follows corporate and regulatory policy rather than any single military alliance. Every staff college now teaches the same line: use it, but never depend on only it.
Civilian lesson: for EmComm groups, a consumer satellite terminal plus a battery bank is the fastest way to restore internet to a shelter or a net control station. Pair it with HF radio and a local mesh so the loss of one path never kills the whole net.
4. Smartphones and ATAK Tactical Software
The US Army once issued every dismounted leader a custom Defence Advanced GPS Receiver the size of a brick. Today many carry a Samsung Galaxy S series phone in a Juggernaut case running the Android Tactical Assault Kit, ATAK for short.
ATAK began at the Air Force Research Laboratory as mapping software for civilian Android phones. It overlays blue force positions, red force reports, drone video, range rings, medevac routes and chat on offline maps, and it talks to radios through plug ins. The Army Nett Warrior program, the Marine Corps and partner forces across NATO field it on the Samsung Tactical Edition phones with locked bootloaders. The software is government furnished, free to US government users, with a civilian build that hikers, search teams and even wildland firefighters use legally.
Why phones won:
Screens, touch response, batteries and processors improve yearly on civilian money, while custom military computers improve on defence budget cycles
Every 19 year old recruit already knows the interface, which cuts training to hours
One device replaces a GPS, a map case, a laser rangefinder display, a drone screen and a messenger
Civilian accessories, from cases to battery packs to vehicle mounts, cost a tenth of mil spec equivalents
The hardening is honest work rather than magic. Tactical phones add physical cases, locked firmware, VPN tunnels, disabled radios and sideloaded rather than store installed apps. They still hate water immersion, extreme cold kills their batteries like any phone, and screen glow needs discipline at night.
Civilian lesson: a phone with downloaded offline maps, a battery case and a cheap Bluetooth GPS puck is 80 percent of a military navigation rig. Hams running APRS alongside ATAK style mapping get the best of both worlds, RF position reports plus rich local maps.
5. Xbox Controllers Driving Submarines and Drones
In 2018 the US Navy replaced the 38,000 dollar custom joystick controlling the photonic masts on Virginia class submarines with an off the shelf Xbox 360 controller costing about 30 dollars. Junior sailors who grew up on consoles learned the control in hours instead of weeks, and spares came from any electronics store rather than a sole source defence contractor.
The pattern repeated everywhere. Military drone operators fly many small UAVs with modified Xbox and PlayStation style controllers. Bomb disposal robots, camera turrets and simulator stations use the same gamepads. The ergonomics argument is simple: Microsoft and Sony spent billions on human factors testing across hundreds of millions of users. No defence lab can match that sample size.
Civilian lesson: if a gamepad is good enough to raise a submarine mast, it is good enough for your antenna rotator, your camera slider or your club robot project. Standardize on one controller family so any member can operate any station.
An Xbox 360 controller of the family the US Navy adopted for Virginia class submarine photonics masts. Photo by Evan-Amos, Wikimedia Commons.
6. HoloLens Augmented Reality and the IVAS Goggles
Microsoft built the HoloLens 2 as an enterprise product for factory technicians and surgeons. The US Army is turning it into the Integrated Visual Augmentation System, IVAS, a ruggedized mixed reality goggle meant to show soldiers navigation cues, friendly positions, weapon sights and drone feeds inside their field of view.
The program, worth up to 21.9 billion dollars over its lifetime, stumbled through early versions with bulk, rain failures and nausea complaints, then pivoted to the slimmer IVAS Next and Anduril partnerships. Whatever the final form, the direction is fixed. The display, the hand tracking and the developer ecosystem come from the civilian product. The Army adds ballistic protection, night vision fusion and tactical software.
Civilian lesson: augmented reality maintenance overlays that guide a soldier through a generator repair will soon guide a homeowner through a furnace repair. Watch this program for where consumer AR glasses go next.
7. iPads in Cockpits and Command Posts
The paper kneeboard, a clipboard strapped to a pilot thigh holding approach plates and checklists, survived for a century. The iPad killed it in about five years.
Starting around 2011, airlines won FAA approval for iPads as electronic flight bags, and air forces followed fast. USAF Air Mobility Command fields tens of thousands of iPads with the ForeFlight style pubs, weight and balance tools and aeromedical apps. Army helicopter crews, artillery fire direction centers and logistics yards use tablets for manuals, maintenance forms and cargo tracking. A 600 dollar tablet replaced flight bags weighing 40 pounds and printing budgets measured in millions.
Why tablets stuck: the civilian display supply chain delivers brighter, lighter and sharper screens every year, app updates arrive without a depot visit, and a cracked screen swaps in minutes. The military adds encrypted containers, mobile device management and rugged cases, but the glass and silicon stay civilian.
Civilian lesson: your flight bag, go box or field binder can follow the same path. One tablet with offline manuals, band plans, repeater lists and PDF checklists, backed by a second cheap tablet in a sealed bag, beats a shelf of binders.
8. GoPro Action Cameras on Helmets
GoPro built a waterproof vitamin sized camera for surfers in 2004. Within a decade it was standard helmet equipment for infantry, special forces, combat engineers and military working dog handlers.
The uses go far beyond publicity clips. After action review of room entries, bridge demolition documentation, vehicle recovery training, parachute jump debriefs and combat engineer route surveys all run on action camera footage. Investigators reconstruct incidents frame by frame. Instructors build training libraries from real operations instead of staged demos. The camera costs a few hundred dollars, mounts to any rail or helmet, survives mud and rain, and any private can operate it without instruction.
The security cost is real. Geotagged footage, visible unit patches and background landmarks have compromised operations repeatedly, and several forces now restrict recording on missions. The camera stays, the discipline around it grows.
Civilian lesson: mount one on your tower climbing helmet, your Field Day mast raising or your SOTA activation. You'll spot rigging mistakes you never felt in real time, and the club training library will thank you.
9. Wristwatches From Trenches to G-Shocks and Garmins
The wristwatch itself is a civilian product the military adopted. In 1900 gentlemen carried pocket watches and strapping a watch to a wrist looked eccentric. By 1917 artillery officers, pilots and infantry leaders needed hands free time for synchronized barrages, and the trench watch, a pocket watch movement in a wrist case with a shrapnel guard, became standard issue. The entire modern watch industry pivoted to the wrist because soldiers bought them.
The pattern repeated twice. Casio built the G-Shock in 1983 for construction workers and bikers who smashed normal watches, engineering a floating module in urethane that survives 10 meter drops and 200 meter water resistance. Soldiers adopted it en masse because it survives what issues watches don't, at a price a private can afford to replace. Then Garmin built GPS sports watches for runners and hikers, and special operations units, pilots and forward observers adopted Fenix and Tactix models for navigation, barometric altitude, night vision compatible backlights and week long batteries.
Civilian lesson: a 100 dollar G-Shock plus a basic GPS watch covers timing, navigation backup and ruggedness for any field hobby. Change the battery on schedule and learn th

https://hamradio.my/civilian-products-used-by-military/

#civilianproducts #modern #military #toyotawar #toyotahilux #adoptionmeans #ukraine #dji #generalsbuy

#rss #feeds #linksharing #qwibbit

From Battlefield to Backyard: 12 Military Products Civilians Use ...

2026-11-03T17:16:48Z

From Battlefield to Backyard: 12 Military Products Civilians Use Every Day

TL;DR: Your house is full of veterans. GPS, the internet, the microwave oven, the Jeep, duct tape, aviator sunglasses, the T-shirt, cargo pants, paracord, canned food, night vision and weather radar all started in uniform. Twelve products inside with the history, the engineering reason each crossed over, and surplus buying tips.
What Military to Civilian Adoption Means
Military to civilian adoption means a product developed for war finds a second life on the high street, a transfer economists call spin off and soldiers call Tuesday, because the surplus store has sold this stuff for a century.
Look around your house and count the veterans. The phone that navigated you here listens to atomic clocks on satellites the US Air Force flies. The page you're reading traveled a network built to survive nuclear war. The lunch you reheated used a magnetron descended from radar sets that spotted bombers. Your sunglasses, your T-shirt, your cargo pants, the tape in your drawer and the cord on your backpack all did military service first. This article covers twelve of them in depth, with the history, the engineering reason each one crossed over, and what to look for when you buy the surplus or civilian version.
Why Armies Invent Things Shoppers Love
Three mechanisms turn war budgets into household goods.
Requirements without compromise. A military buyer writes requirements civilians never would, works after total immersion, survives a parachute drop, feeds an army without refrigeration. Meeting those extremes usually means inventing a new material, a new process or a new device. Once invented, the cost falls and the civilian version follows.
Scale that collapses prices. The US military bought GPS receivers by the hundred thousand, paracord by the million feet, duct tape by the train car. That volume pays for the factories, the tooling and the learning curve. Civilian buyers then inherit mature supply chains at commodity prices.
Surplus that seeds markets. Every war ends with warehouses of gear sold for cents. Jeep created the civilian 4×4 market from surplus Willys trucks. Night vision companies, tactical clothing brands and the entire paracord bracelet industry grew the same way, from cheap ex military stock that taught civilians what the gear could do.
The honest caveat runs through every section below. Military grade often means built to a price for a narrow job, not built to last forever. Surplus varies from unissued treasure to worn out junk. Buy with the checklist at the end, not with romance.
A 1943 Willys MB, the military Jeep that became the template for every civilian 4×4. Photo: Wikimedia Commons, public domain.
1. GPS: From Missile Guidance to Every Pocket
The Global Positioning System is the largest military gift to civilian life ever given, and it was given almost by accident.
The US Navy built Transit in the 1960s so submarines could fix their position from satellite Doppler shifts. The Air Force and the Pentagon merged the rival programs into NAVSTAR GPS in 1973, launching the first satellite in 1978 and declaring full operation in 1995. The design goal was guiding weapons and troops, with a deliberately degraded civilian signal called Selective Availability that capped public accuracy around 100 meters.
Two decisions changed the world. Ronald Reagan ordered the system opened to civilian airliners after Korean Air Lines Flight 007 was shot down in 1983 for straying off course. Bill Clinton switched Selective Availability off in 2000, and overnight civilian accuracy jumped from 100 meters to about 5. Phone makers then shrank receivers to chips costing a few dollars. Today more than 30 GPS satellites plus Russian, European and Chinese constellations serve around 6 billion receivers, in phones, cars, tractors, wristwatches and the timing circuits of power grids and cell towers.
Key facts that explain the crossover:
Atomic clocks on each satellite, stable to nanoseconds, which is why your bank timestamp and your map share one clock
L1 civilian signal at 1575.42 MHz, free worldwide, no subscription, no account
Assisted GPS, where cell towers feed orbital data so a cold phone locks in seconds instead of minutes
Disciplined oscillators, where a 20 dollar GPS module steers a ham shack frequency standard to atomic accuracy
Buying tip: any modern phone is a GPS receiver. But a dedicated Bluetooth GPS puck plus offline maps survives dead zones and dead phone batteries. For shack frequency discipline, a surplus timing GPS module beats a free running oscillator at any price.
A GPS Block IIF satellite. The same constellation guides missiles and lost hikers alike. Illustration: US Air Force, public domain.
2. The Internet: A Network Built to Survive Nuclear War
The network carrying this article began as ARPANET, funded by the Pentagon Advanced Research Projects Agency and switched on in 1969 with four university nodes. The goal was resource sharing between research computers, hardened by packet switching so messages could route around destroyed links.
The military fingerprints run deep. TCP/IP, the language of the net, came from DARPA funded work by Vint Cerf and Bob Kahn. Email headers, domain names and the first spam filters all grew in defence funded labs. The .mil domain still sits beside .com in the root. When the National Science Foundation lifted the commercial ban in the early 1990s, the same protocols carried web shops instead of research files, and the rest is the economy you live in.
What civilians inherited without noticing:
Packet switching, which makes a video call survive a severed cable by rerouting mid sentence
End to end encryption concepts from military COMSEC, now the padlock on every bank login
GPS time sync from section 1 above, which stamps every trade, log and certificate
The culture of open standards, because the Pentagon demanded vendors interoperate
Buying tip: none needed, you already own it. But run your home net like the designers intended, with more than one path. A phone hotspot backing up fixed broadband is packet switching thinking applied to bills.
3. The Microwave Oven: Melted Chocolate Near a Radar Set
In 1945 Raytheon engineer Percy Spencer stood near an active magnetron during radar testing and felt the chocolate bar in his pocket melt. Where others saw a ruined snack, Spencer saw a heating method. He aimed the tube at popcorn, then at an egg that exploded in a colleague face, and filed the patents for microwave cooking.
The first commercial Radarange of 1947 stood 1.8 meters tall, weighed 340 kilograms, cost about 5,000 dollars and needed water cooling. It heated meals in ship galleys and railway canteens. Two decades of magnetron cost cutting, driven by defence radar volume, shrank it to the countertop box of the late 1960s, and the 1,000 watt home oven became the fastest adopted appliance in history.
Why it crossed over so completely:
Magnetrons were already mass produced for radar, so the expensive part kept getting cheaper
Speed fit postwar life, since reheating in minutes matched women entering the workforce and frozen food going mainstream
Safety engineering, the interlocked door and Faraday cage, made megawatts of RF boring to regulators
Buying tip: the same Faraday cage principle shields your shack. A dead microwave with the cord cut makes an excellent RF tight box for storing spare handhelds against lightning induced surges, and the turntable motor runs many slow antenna projects.
4. The Jeep: A Quarter Ton Truck That Invented the SUV
In 1940 the US Army asked for a quarter ton 4×4 reconnaissance truck, and Willys Overland plus Ford delivered the Willys MB and Ford GPW. About 640,000 rolled out, serving on every front as gun platforms, ambulances, rail tugs and generals runabouts. General Eisenhower called it one of the three tools that won the war, beside the Dakota and the landing craft.
Willys saw the peace coming and trademarked the shape, launching the CJ-2A Civilian Jeep in 1945, the first mass produced civilian 4×4. Farmers, surveyors, foresters and hunters bought the same ladder frame, live axles and flathead engine the Army trusted. Land Rover copied the formula in Britain in 1948, Toyota followed with the Land Cruiser, and the SUV segment that now dominates world car sales descends directly from that quarter ton requirement.
Engineering traits that survived into your driveway:
Short wheelbase and flat panels, easy to repair with hand tools and a welder
Part time 4×4 with low range, still the layout of every serious off roader
PTO and flat fenders that doubled as workbenches, a tradition ending in modern tailgate tables
Buying tip: surplus Jeeps are mostly gone, but the buying logic holds for any ex military 4×4. Check for straight frames, parts commonality with civilian models, and service history. A truck every village mechanic knows beats an exotic import every time.
5. Duct Tape: Sealing Ammo Cans, Fixing Everything Else
In 1943 Vesta Stoudt, an ordnance plant worker with two sons in the Navy, noticed ammo boxes sealed with paper tape and wax failed in damp Pacific warehouses. She wrote to President Roosevelt proposing a waterproof cloth tape, the Army tested it, and Johnson and Johnson division Permacel shipped olive drab tape that sealed cases, repaired tents, fixed boots and silenced rattling gear.
Soldiers called it hundred mile an hour tape and brought the habit home. The civilian silver version arrived as heating duct sealing, which gave the name, and the uses exploded from stagecraft to spaceflight. Apollo 17 astronauts taped a fender extension from lunar maps with it, the most famous field repair ever made.
What makes it a military product at heart:
Cloth reinforced polyethylene, strong enough to lash, flexible enough to wrap
Aggressive adhesive that bonds dirty metal, canvas and skin in rain
Tears by hand along the weave, so no tools needed in the dark
Buying tip: buy named brand tape with listed thickness, not thin bargain rolls that shred. Olive drab or black looks better on radio gear and avoids the shiny silver giveaway at field sites. Store rolls sealed, since the adhesive dries with age.
Government issue duct tape, direct descendant of the ammo box sealer. Photo: Wikimedia Commons.
6. Aviator Sunglasses: Goggles for Bomber Pilots
High altitude bomber crews in the 1930s suffered headaches and nausea from blue sky glare above the clouds. The US Army Air Corps asked Bausch and Lomb for protection, and in 1936 the Ray-Ban Aviator shipped with green G-15 lenses that cut visible light while preserving instrument readability, in a thin gold frame light enough for oxygen masks.
General MacArthur wore his on the cover of every newspaper, and returning airmen kept wearing theirs with civilian clothes. The teardrop shape that covered a pilot peripheral vision became the default cool of the next eighty years, from Hunter S Thompson to every highway patrol film. The Wayfarer, the Clubmaster and half the sunglass wall trace back to that one contract.
Buying tip: real G-15 style green glass lenses still offer the best contrast for driving and flying. For radio field work, brown gradient lenses read LCD screens better than flat grey.
7. The T-Shirt: A Navy Undershirt That Conquered the World
The US Navy issued the first crew neck short sleeve undershirts in 1913, plain white cotton worn under jumpers in engine rooms and tropics. Cheap, washable and cool, it spread through the Army in World War I and came home on millions of backs.
Hollywood finished the job. Brando in A Streetcar Named Desire and Dean in Rebel Without a Cause wore the undershirt as the shirt, and the blank cotton chest became the cheapest billboard ever sold, for bands, causes, companies and callsigns. Your club shirt with the repeater frequency is the direct descendant of a Navy skivvy.
Buying tip: for field events, heavyweight cotton or cotton poly blends survive tower work that thin fashion shirts don't. Print the event, the date and the talk in frequency, and the shirt becomes a QSL card people wear.
8. Cargo Pants, Field Jackets and Trench Coats: Uniforms on the High Street
Three garments marched off base and never went back.
The trench coat came first. Thomas Burberry waterproof gabardine officer coats of World War I, with epaulettes, D rings for grenades and storm flaps, returned as civilian rainwear and never left film noir. The M-65 field jacket of 1965, with its hood in the collar and brass zips, went from Vietnam to protest marches to fashion week. Cargo pants, with bellows pockets sized for maps and magazines, went from paratroopers to schoolyards to every maker workbench.
Why military clothing colonizes wardrobes:
Pockets sized for real objects, not decoration, which civilians discover the first time they carry tools
Fabrics specified for abrasion and washing, not showroom drape
Cuts that layer, since soldiers dress for weather that changes by the hour
Buying tip: genuine surplus field jackets and cargo trousers cost less than fashion copies and last longer. Check seams, zips and elastic before buying, and size for layers rather than the mirror.
9. Paracord 550: Parachute Lines for Everyday Survival
Type III parachute cord, rated 550 pounds, held American paratroopers under silk canopies from World War II onward. Riggers and troopers quickly found the seven inner yarns plus outer sheath repaired boot laces, guyed shelters, lashed litters and fished tripwires. It came home in kit bags and entered the civilian world through surplus bins and survival manuals.
Today it lashes antennas, hangs food bags, sews field repairs and weaves the survival bracelets sold at every hamfest. The spec matters, because the name gets abused. Real 550 cord has seven twisted inner strands, a 32 strand woven sheath, and genuine Mil-C-5040 or commercial equivalent marking. Much of what craft stores sell is single strand filler that fails at half the rating.
Ham uses worth learning:
Doublet and dipole guys that survive storms without stretching like nylon rope
Lanyards and pull cords sized to the exact load table, not guesswork
Field splicing and whipping, since the inner yarns sew and the sheath chafes through first by design
Buying tip: buy from cordage dealers who publish break tests, avoid unmarked craft cord for anything overhead, and learn three knots, bowline, taut line hitch and prusik, before buying a fourth color.
10. Canned Food: Napoleon Prize That Fed the World
In 1795 the French Directory offered 12,000 francs for a way to preserve food for armies on the march. Nicolas Appert, a Paris confectioner, won it in 1809 by sealing food in glass jars and boiling them, and the Royal Navy adopted tin canisters within a few years. Canned rations fed every army of the next two centuries and every pantry beside them.
The military kept pushing the science. World War II drove volume and quality control, Vietnam produced the first retort pouch experiments, and the 1981 Meal Ready to Eat program perfected the full menu in a bag with flameless heater. Civilian campers, preppers and disaster agencies eat the direct results, from Mountain House pouches to tuna sachets to the retort curry on Malaysian shelves.
Buying tip: rotate stock like a quartermaster. First in first out, date every box in marker, and store what you actually eat so rotation happens by appetite rather than discipline.
11. Night Vision: Owning the Night, Then Selling It
The US Army fielded starlight scopes in Vietnam that amplified moonlight thousands of times, turning night patrols into daylight business. Each generation shrank the tubes, sharpened the image and cut the price, from truck mounted infrared searchlights to helmet goggles to weapon sights.
Surplus and declassified tubes seeded the civilian market. Hunters scan fields for hogs, boaters read unlit channels, security teams watch perimeters, and firefighters see through smoke with thermal cameras that descend from the same programs. Phone clip on thermal imagers now find wall leaks and sick cattle for a few hundred dollars, a capability that cost a division budget in 1970.
Limits worth respecting:
Civilian tubes and digital sensors lag military issue by a generation or more in range and clarity
Export controls follow the good stuff across borders, so check the law before traveling with it
Light discipline still applies, since an IR illuminat

https://hamradio.my/military-products-civilians-use-every-day/

#military #civilianadoption #surplus #means #gps #jeep #airforce #willys #everypocket

#rss #feeds #linksharing #qwibbit

Libreon: A Private Offline Office Suite for Android with No Ads ...

2026-11-03T16:36:54Z

Libreon: A Private Offline Office Suite for Android with No Ads and No Accounts

TL;DR: Libreon is a free GPL-3.0 Android office suite that opens, edits and saves Word, Excel and PowerPoint files plus PDF and text with no ads, no account, no trackers and no internet permission at all. Built on LibreOfficeKit and PDFium in Kotlin and Jetpack Compose, about 115 MB per ABI, Android 12 and newer. Full install guide, privacy design, architecture tour and honest beta limits inside.
What Libreon Is
Libreon offline office is a private, fully offline office suite for Android that opens, edits and saves Word, Excel and PowerPoint files, views and signs PDF documents, and handles plain text and Markdown, with no ads, no subscriptions, no trackers, no AI features and no internet permission at all.
That one sentence carries the whole pitch, so unpack it slowly. Libreon is a native Android app written in Kotlin with Jetpack Compose, currently in public beta at version 0.9.0 beta 5. It is free software under GNU GPL-3.0 or later. The parts that actually parse your documents aren’t new code. They are LibreOffice core reached through LibreOfficeKit with the Collabora Online web front end running locally in a WebView, and PDFium, the same PDF engine that ships inside Chrome, reached through a small JNI bridge. Around those engines sits a brand new shell: tabs, recents, file search, find in document, word counts, light and dark themes, and a recovery system that keeps unsaved changes alive even when Android kills the app in the background.
The project lives at https://github.com/jainiakhil/Libreon by developer Akhil Jain, with 35 commits and five beta releases in about two days at the time of writing, which tells you how fast the feedback loop is running. The tagline on the repository is three words long: Work freely. After a decade of office apps that treat a phone like a billboard with a document viewer attached, three words feel about right.
Libreon, Work freely. A private offline office suite for Android with no ads, no accounts and no internet permission. Artwork: Akhil Jain, Libreon project, GPL-3.0.
Why Your Phone Needs an Office App That Cannot Phone Home
Every Android user knows the drill the Libreon README opens with, because every one of us has lived it. You tap a Word file a club member sent you. A full screen ad plays. A Go Premium popup slides up. A sign in nudge follows. Then an AI assistant offers to summarise the document you haven’t even read yet. The document appears with a banner ad parked over the bottom third. To open a file you already own, you paid with attention, data and patience.
This didn’t happen by accident. It’s the business model. Desktop office software used to be a product you bought once. Mobile office software is a service you rent forever, monetised through subscriptions, advertising inventory, analytics pipelines and upsold AI credits. The document is the bait. You are the inventory. Microsoft 365 wants the subscription, Google Docs wants the account and the cloud corpus, WPS Office wants the ad impressions. Each one is excellent at editing documents, because each one has to be good enough that you tolerate the price around it.
Radio amateurs have a special reason to dislike this deal. Field work happens where the network doesn’t. A summit, a rural EmComm deployment, a ship at sea, a border crossing with no roaming, a concrete basement club meeting with one bar of signal. An office app that needs to phone a server to open a local file isn’t a tool, it’s a liability. The same instinct that makes hams keep paper duplicates of critical logs, run offline loggers for Field Day, and prefer tools that work with the cell network switched off applies to documents. Activation checklists, EmComm ICS forms, equipment inventories, antenna build notes, QSL card mail merges. These live as docx, xlsx and PDF files, and they need to open on a mountaintop.
Libreon answers with a clean room design and a permission manifest anyone can audit. Clean room here means what it says in the README. No code, resources or artwork were taken from any existing app. Existing apps were studied only to learn what features people expect. Everything is written from scratch or comes from properly licensed open source projects with a full license list in docs and on the in-app open source licenses screen. And the network story isn’t a policy promise, it’s an Android enforcement guarantee. The merged manifest explicitly removes the INTERNET permission, a CI check fails the build if any uses-permission appears, cleartext traffic is off, and a network security config denies everything. On Android that is checkable, not marketing. Open the manifest and the permission is absent, so the app physically can’t upload your documents, your signatures or your keystrokes no matter what a future update might want. The only web links in the whole app sit at the bottom of Settings, the GitHub page and a buy me a coffee banner, and they open in your browser only when you tap them.
There is an irony the author discloses upfront and deserves credit for. Libreon ships no AI inside, yet it was built with two AI coding assistants, Claude Code and Google Antigravity, while the developer designed, reviewed and tested on real phones. The AI wrote code, it didn’t move into the app. Because every line is public and the app can’t reach the network, you don’t have to take his word for any of it.
What Libreon Actually Does
Three engines cover the whole job. Office documents go through LibreOffice, PDF through PDFium, and text through a custom editor. Each one is worth understanding on its own terms.
Office documents open, edit and save in the formats people actually exchange. Word, Excel and PowerPoint in the modern OOXML set, docx, xlsx and pptx, plus the legacy binary set, doc, xls and ppt. OpenDocument in all six flavours, odt, ods and odp plus flat fodt, fods and fodp, and RTF. Apple iWork files, Pages, Numbers and Keynote, open read only through the libetonyek import filters the engine already carried. Find in document with match counts works across Office files through the front end search service, with next and previous moving the selection, and Word documents get a live word count line under the page from Writer own counter.
PDF covers the daily document grind that never needed a word processor. Smooth viewing with pinch zoom, page thumbnails and text search. Real AcroForm filling with text fields, checkboxes, radio buttons and dropdowns written as genuine field values with regenerated appearance streams, so Adobe Reader and every other viewer shows them. Flat forms with no fields get a Fill tool that free places text plus check and cross marks, the WPS pattern, committed as FreeText and stamp annotations. Signing uses drawn or imported signatures stored only on the device, placed in a resizable movable box, and flattened into page content by default with keep as annotation available as an option. Highlighting works the way readers expect. Long press a word, drag the handles, pick one of four colours. Highlights are standard Highlight annotations with quad points and the text in Contents, so other PDF apps display them and they survive saving and tab recovery. Copy selected text with a word count in the bar, and count words for the whole document from its text layer.
Text handles the files other office apps refuse to touch. Plain text, Markdown with a locked down no JavaScript preview, CSV with a read only table view plus open as spreadsheet handoff to the Office engine, JSON with pretty print validation, XML, logs, INI and CONF, YAML, subtitles and config files, basically any text MIME. Files up to 2 MB open editable. Between 2 and 50 MB they open read only in a virtualised line list. Above 50 MB the app offers read only head and tail views. Encoding, byte order mark and line endings round trip byte exactly, with BOM first, then strict UTF-8, then a small built in heuristic for UTF-16 without BOM, East Asian multibyte, Windows 1252 and ISO 8859-1. Find and replace, go to line, a live Ln Col word and character status line, word wrap and monospace toggles, and one shared word counting rule with JVM tests that counts Chinese and Japanese characters one each.
Everywhere features tie the three engines together. Tabs keep many documents open at once, with hot sessions holding a loaded engine and parked sessions saving a recovery copy and unloading under a memory budget, one live Office document, up to three live PDFs, unlimited small text buffers. If Android closes the app in the background, unsaved changes are still there when you come back, with a restore banner when the recovery copy is newer than the source. Search on the home screen filters open, recent and recently closed files by name. Light, dark and follow the system themes apply across app, Markdown preview and Office pages live, with a per document light pages option while the app stays dark. A themed launcher icon follows the wallpaper, and screen reader labels cover buttons, tabs, rows and toggles.
Privacy as Architecture, Not a Toggle
Most privacy apps ask you to trust a switch. Libreon asks you to read a manifest, because the guarantee is structural.
No internet permission is the headline, but the surrounding decisions are what make it hold. No accounts, no analytics, no crash reporters, no Play Services, no Firebase. Gradle dependency verification plus an allowlist enforces that nothing network capable sneaks in through a transitive dependency. Settings and signatures live in app private storage with cloud backup excluded, so signatures never ride a backup to a server. Saving writes to a private temp file, fsyncs, streams to the Storage Access Framework URI, verifies the byte count, and only then drops the temp copy. Before overwriting, the app compares the provider size and modification time against a stored fingerprint, and asks before clobbering a file another app changed meanwhile. If you open a file from a share intent with only a transient grant, edits go to the private copy and Save becomes Save as, which is the honest behaviour instead of silently losing work.
The support model follows the same ethic. Libreon is free and the author states it always will be, no ads, no subscriptions, no premium tier, no nagging. The single buy me a coffee banner lives at the very bottom of Settings, added in beta 4, with no popups, no reminders, no held back features. A report a problem link in the same screen opens GitHub Issues in your browser. Ignoring both is completely fine, and the author says so in the README.
For hams this maps directly onto familiar threat models. A club membership spreadsheet with phone numbers, an EmComm deployment roster, an equipment loan ledger, scanned QSL cards with home addresses. These are documents that should never transit a third party server just to be viewed on your own phone. An app that can’t reach the network can’t leak them, can’t train on them, can’t hold them hostage behind a login wall during an outage.
Under the Hood: LibreOfficeKit, PDFium and a Two Process Trick
The architecture document in the repository is unusually candid, and it repays reading, because it explains why Libreon exists in this exact shape and why nobody else built it this way sooner.
Office editing reuses LibreOffice core through LibreOfficeKit, with the Collabora Online browser front end running locally in a WebView. Assets load from the WebViewAssetLoader path, JavaScript talks to native code through a WebMessagePort bridge the project wrote itself, and every other network request is rejected in shouldInterceptRequest. Each Office document runs in its own engine process shown in the app through a shared surface, so a heavy document can’t take the app down. This two process design, option 2 of three considered in the architecture record, was chosen after the team confirmed that on Android each document kit runs LibreOffice main loop itself, so closing a document ends the process. Upstream Collabora Android glue assumes one document per process with its own Activity. Libreon instead hosts OfficeService in a dedicated office process and renders it into the main window through SurfaceControlViewHost, keeping the shared tab bar unchanged. Office to PDF switches detach the surface and keep engine state. Office to Office switches start a fresh engine process in about 2 seconds. The app process itself never dies when an engine crashes. It shows the engine stopped unexpectedly with a Reload button that reopens from the recovery copy.
The build cost explains the 115 MB download. Cross compiling LibreOffice core needs a Linux host with WSL2, tens of gigabytes of disk and about three hours per ABI on 12 cores, plus the Collabora online repository for the front end and Android glue. The team slimmed the engine from 204 MB stripped to 158 MB through feature disables, Os optimisation and compact RELR relocations, compressed the native libraries in the APK, and ships per ABI splits, 115 MB for arm64 and 118 MB for x86_64. The 33 MB ICU locale file is the next trimming candidate. Building without a signing key falls back to the debug key, and releases are signed with a private key kept outside the repository.
PDF takes the opposite path, small and hand built. PDFium from the Chromium 8100 prebuilt plus about fifteen hundred lines of new JNI, one engine for view, fill, sign and save. Tile based rendering into bitmaps with a low resolution pass then high resolution tiles, an LRU cache under the memory budget, all PDFium calls on one dedicated thread per document because PDFium isn’t thread safe. Save uses FPDF SaveAsCopy, incremental when not flattening, always save as new file by default for signed documents. The team fuzzed the JNI, fixed freed page handoffs and unchecked allocations found in a cross model review, and re verified output with pypdf, field values present, flattened signatures in page content, annotation only adding exactly one Stamp, original bytes untouched.
Text is the smallest engine and the most opinionated. Compose BasicTextField with the state based TextFieldState API for built in undo and streaming line indexing for very large files, commonmark-java for Markdown, a home grown RFC 4180 parser for CSV, org.json from the platform for JSON formatting. Markdown preview renders through commonmark-java with GFM tables into HTML shown in a WebView with JavaScript disabled and all network loads blocked. The whole text milestone carries 25 JVM tests, PDF adds 21 more, Office adds 18 for socket framing and engine messages, 64 in total.
Tabs and storage complete the picture. A Room database holds the tab registry, view state per kind, dirty flags, recovery paths and source fingerprints. Recovery copies write after 2 seconds idle for text, every 60 seconds while dirty for Office, and as an edit log replay for PDF annotations. Storage uses the Storage Access Framework only, with persistable grants for picker results and private working copies for transient share intents. No storage permissions at all.
Installing Libreon in Five Minutes
Get the APK from the releases page at https://github.com/jainiakhil/Libreon/releases. Pick arm64-v8a for almost every phone and tablet, or x86_64 for emulators and some Chromebooks. Open the file and allow installing from your browser or file manager when Android asks. It requires Android 12 or newer, which matches minSdk 31, and the download is about 115 MB because a whole office engine lives inside.
First open unpacks engine files, about 2.7 seconds on the emulator, longer through ARM translation. Open a docx from your downloads through the system picker, grant the persistable URI permission, and edit. The Office page follows your theme automatically. Try the per document light pages option in the document menu if you read white page documents at night with the app in dark mode. Open a fillable PDF, tap a field, fill it, add your saved signature, and use save as new file so the original stays pristine. Open a huge log or CSV and flip on the table view. Kill the app from recents with unsaved edits, reopen it, and confirm the restore banner. That single test tells you more about the tab recovery design than any feature list.
Beta testers should read the known gaps before adopting it daily. Printing and exporting Office documents to other formats aren’t avai

https://hamradio.my/libreon-offline-office-suite-android/

#officesuite #offlineoffice #privateoffline #libreon #ads #jetpackcompose #accounts #word #akhiljain #pdf

#rss #feeds #linksharing #qwibbit

NyaTerm: A Modern Remote Terminal Workspace for SSH, SFTP, Serial ...

2026-11-03T15:58:35Z

NyaTerm: A Modern Remote Terminal Workspace for SSH, SFTP, Serial and RDP

TL;DR: NyaTerm is a free MIT-licensed remote terminal workspace built in Rust on the GPUI framework that puts SSH, local shells, Telnet, Serial, RDP and VNC sessions, an SFTP browser with transfer queue, tunnels, OTP codes and encrypted sync into one fast native desktop client for Windows, macOS and Linux. About 2,200 stars and 4,100 commits at the time of writing, with session import from Xshell, MobaXterm and WindTerm. Full install guide, feature tour, architecture breakdown and honest limits inside.
What NyaTerm Is
NyaTerm remote terminal workspace is a native desktop client for SSH-centred operations and mixed terminal workflows, written in Rust on the GPUI interface framework, that combines terminal sessions, file transfer, remote monitoring and encrypted configuration sync in a single application window.
Unpack that in plain terms. Most engineers live in three or four separate tools: one SSH client for servers, another program for file transfer, a serial console for routers and embedded boards, a remote desktop viewer for Windows boxes, and a password manager plus an authenticator app holding the keys to all of it. NyaTerm folds that whole desk into one workspace with tabs, horizontal and vertical splits, side panels for files, network, security, sync and notes, and an AI assistant panel for generating commands and explaining output. It isn’t a shell replacement, the project says so explicitly. It connects to remote shells, local shells, Telnet endpoints and serial devices, and leaves the shells themselves alone.
The project lives at https://github.com/nyakang/nyaterm, MIT licensed, with installers and documentation at https://nyaterm.app. The numbers signal a healthy project: roughly 2,200 stars, 207 forks, 66 open issues against more than 4,100 commits, plus a Discord community, a WeChat group and a contributors wall. Downloads come from the site or the releases page, with Windows exe and portable zip, macOS dmg plus a Homebrew cask, and Linux deb, AppImage and rpm plus a community AUR package.
NyaTerm main workspace in dark theme with tabbed sessions, activity bars and file panels. Screenshot: NyaTerm docs site, MIT licensed.
Why Another Terminal Client
The terminal client shelf looks crowded until you sort it by architecture and incentives, and then the gap NyaTerm fills becomes obvious.
PuTTY is the old reliable. Free, tiny, and essentially frozen in design since the early 2000s. It speaks SSH, Telnet and serial well, stores sessions, and asks nothing of your GPU or your wallet. What it never became is a workspace. One window per connection, no tabs in the classic build, no file transfer beyond PSCP on the command line, no sync, no scripting worth mentioning. For one router it is perfect. For forty servers it is window-management Tetris.
MobaXterm went the other direction and won a generation of Windows sysadmins. Tabs, embedded X server, SFTP browser beside the terminal, serial, RDP and VNC in one window, macros and a portable edition. The catch is the licence. The Home edition is free with session limits, the Professional edition is per-user paid software, and the whole thing is closed source from one vendor. When the vendor roadmap drifts from your needs, your only vote is your wallet.
WindTerm showed what a free integrated workspace could feel like, with a fast terminal, SFTP, serial, and a famous spring-latch aesthetic, then stalled. Its repository went quiet, binaries lagged, and users who built workflows on it started asking what happens when an unaudited closed-source-feeling binary stops shipping updates. The trust question matters for a tool that holds your SSH keys.
Tabby answered with open source and plugins, but on Electron. A browser engine per window buys easy cross-platform UI at the price of hundreds of megabytes of RAM and a startup time you can feel. On a modern desktop that is tolerable. On a field laptop from 2018 with sixteen tabs open it isn’t.
Then Zed proved a third path. Its GPUI framework renders native-feeling interfaces on the GPU with Rust performance and startup times measured in milliseconds, not seconds. NyaTerm takes that bet for the terminal world: a Rust workspace of a dozen crates, one native client per OS, no browser engine, no Java runtime, no Electron. The credits file names the lineage honestly: WindTerm inspired the design, Tabby supplied design ideas, Alacritty contributed terminal emulation components, and russh provides the Rust SSH implementation. Standing on those shoulders beats reinventing any of them.
Sessions and Workspace
The session model is the heart of the app, so start there. Six session types cover the realistic remote-access spread: SSH for Linux and Unix servers, local terminal for the machine under your fingers, Telnet for legacy lab gear, serial for routers, boards and embedded devices, and RDP plus VNC panes for graphical remote desktops. Serial sessions expose port, baud rate, data bits, parity, stop bits and a backspace mode selector for Ctrl-H versus DEL, the kind of detail that decides whether a console session is usable or garbage. VNC currently handles direct TCP with no-auth or classic VNC auth, Raw, ZRLE, Tight and Tight JPEG encodings, window scaling, bounded reconnects and Latin-1 clipboard exchange.
Everything lives in a multi-tab workspace with horizontal and vertical splits, tab drag docking and layout restoration, so a router console, two server shells and an SFTP browser can share one window as one task. An activity bar on each side hosts panels for file explorer, network, security and auth, sync and backup, AI assistant, notes, saved connections, active sessions, command history, asset monitoring, resource monitoring, GPU monitor, process manager and Docker manager. Saved connections carry folders, icons, metadata, duplication and reconnect, with a command palette and session quick switcher for jumping between dozens of open sessions by typing. Temporary SSH links open one-off connections from a pasted ssh URL or ssh command line without polluting the saved list. Input sync groups broadcast typed input to several sessions at once, which is how you restart the same service on eight nodes without scripting. Child windows keep settings, connection wizards, quick-command editing and remote file editing out of the way, and tray support with minimize-to-tray suits a jump-box workflow.
Two thoughtful touches stand out. First, background image customization with cover, contain, stretch and tile sizing plus content opacity control, because people who live in a terminal care what it looks like. Second, notes with tree navigation, autosave and sync-aware persistence, so runbooks sit next to the sessions they describe instead of in a separate wiki tab you forget to open during an outage.
Terminal Experience
A pretty frame means nothing if the terminal itself is weak, and this is where the Alacritty DNA shows. Terminal search covers result navigation, history, copy and paste, context menus and selected-text actions. Command history offers fuzzy suggestions with length filters and stays quiet inside interactive programs. Unicode grapheme rendering handles emoji, combining marks and zero-width-joiner sequences instead of drawing tofu boxes. An optional gutter shows line numbers and timestamps, action links turn IPv4 addresses, host-colon-port strings and archive filenames clickable, and keyword highlighting ships with expanded presets, custom rules and JSON import.
Daily ergonomics get the details right. Terminal zoom, workspace padding, font weight controls, native IME support for CJK input, image path paste handling, clear-terminal action, online search and translation straight from selected text, session recording for audits, configurable scrollback, SSH keep-alive so NAT gateways stop eating idle sessions, and large-output protection so a runaway cat of a gigabyte log doesn’t freeze the UI. Zmodem transfers run directly from the terminal and land in the same transfer queue as SFTP jobs. A confirmation dialog guards closing all sessions, and keyboard shortcuts are customizable including the Telnet and serial backspace behaviour.
SFTP and File Workflows
File work is where all-in-one clients earn their keep, and NyaTerm treats it as a first-class pane rather than a bolt-on. The built-in SFTP explorer rides beside the SSH session with upload, download, rename, move, delete, properties, new file and folder creation, and OpenSSH-compatible symlink actions. Folder upload and download, multi-select, an editable path bar, and manual or automatic sync with the terminal working directory keep the two views coherent. The transfer queue shows speeds and supports pause, resume, cancel, retry, duplicate-target handling, timestamp preservation and configurable concurrency, with channel concurrency limiting and automatic retry on transient channel-open failures.
Two workflows deserve special mention. Remote files open in your local editor and saved changes upload back through a watcher-driven auto-upload flow with content fingerprinting, so only real content changes trigger re-upload instead of every save event. And a multi-tab read-only preview handles text, JSON, Markdown, CSV and TSV, images and PDF with per-tab session labels, zoom, rotate and fit controls plus paged PDF rendering, decoded off the render thread with per-format size limits so one monster file can’t wedge the interface. External drag and drop on Windows and reveal-in-file-manager for completed Zmodem transfers round out the edges.
Security, Authentication and Networking
A terminal workspace is a crown-jewels application, so read this section as an audit checklist. Authentication covers passwords, private keys, host-key verification and encrypted local persistence. Credentials support regex-based terminal auto-fill for those login banners that never match a standard prompt. OTP management handles TOTP and HOTP with QR import and SSH auto-fill, collapsing the authenticator app into the session. Per-connection SSH algorithm preferences offer Compatible, Secure and Custom modes with security-risk labels on key exchange, ciphers, MACs and host keys, so legacy gear gets a conscious downgrade instead of a silent one. Proxies span SOCKS5, HTTP and ProxyCommand, jump hosts include cycle prevention, tunnels cover local, remote and dynamic forwarding, and SSH X11 forwarding serves environments with a local X server.
Hygiene controls complete the picture. Screen lock plus idle app lock, a master password guarding storage, sync, backup, import and export, diagnostics settings with local log management and exportable bundles. Sync itself demands the master password before anything moves.
Sync, Backup and Migration
Configuration portability decides whether a tool survives a laptop replacement, and NyaTerm over-invests here in a good way. Encrypted cloud sync and backup run through WebDAV, S3-compatible storage and GitHub Gist, all gated on the master password, with startup sync checks, debounced auto-push after local changes, scheduled backups with retention, manual test, push, pull and backup actions, remote restore and snapshot-level conflict resolution. The full-app encrypted .nya format exports everything for portable moves.
Migration paths acknowledge reality. Session import reads Xshell, MobaXterm and WindTerm definitions plus NyaTerm JSON, which is exactly the on-ramp a team needs when standardizing. One caution from the docs: moving from 1.x to 2.0 on Windows portable and old MSI installs needs a manual upgrade with a full data and key backup first, replacing the complete portable payload while preserving the data directory, or uninstalling the old MSI before installing the NSIS build. Replacing only the main executable is insufficient.
AI Assistant Where It Is Useful
The AI panel splits into two modes with a clear risk model. Ask mode answers one-off questions: generate a command, explain selected output, analyze an error. Agent mode runs multi-step work in an observe-decide-run loop against the active terminal, with inline capture of a configurable number of output lines, session mentions via at-sign to pull other sessions into context, and structured command cards carrying risk levels, execution controls and optional save-to-quick-command. Approved command execution tools gate the dangerous steps, with a separate final-answer step after terminal work completes. Providers include built-ins, manual model entries, credential groups and custom OpenAI-compatible endpoints, and recent-output actions explain the latest terminal output without manual copying. High-impact commands hit approval gates with safer alternatives suggested. It is assistance with a brake pedal, not autopilot.
Under the Hood: Rust, GPUI and Twelve Crates
The repository structure reads like a textbook Rust workspace. Nyaterm-app holds the native entry point and assets, nyaterm-desktop composes the application and feature state, nyaterm-core keeps UI-independent domain models and policies, nyaterm-terminal owns the state machine and parsing with nyaterm-terminal-gpui handling layout, input and painting, nyaterm-transport covers SSH, PTY, SFTP, Telnet, serial and transfers, nyaterm-store handles persistence and database compatibility, nyaterm-ui shares controls and theme integration, nyaterm-remote-desktop plus the isolated IronRDP and VNC helper processes cover graphical sessions over typed IPC, and nyaterm-otp implements one-time passwords. The docs site is Docusaurus in the same monorepo. Contributors build with a stable Rust toolchain via rustup plus platform GPUI dependencies, then run the app crate directly with cargo. The AGENTS.md and CLAUDE.md files in the repo root show an AI-assisted contribution workflow, which matches the fast commit cadence.
RDP and VNC isolation deserves a note. Running the IronRDP protocol helper and the VNC helper in separate processes means a protocol parser bug crashes a helper, not your twelve-tab workspace. That’s the same blast-radius thinking behind browser sandboxing, applied to remote desktop.
Installing NyaTerm in Minutes
Grab the latest build from https://nyaterm.app or the releases page. Windows takes the exe installer or the portable zip where data lives in the adjacent data directory. MacOS takes the dmg dragged into Applications or the Homebrew cask from the nyakang tap with brew install. Linux takes deb, AppImage or rpm, and Arch users have the community nyaterm-bin package in the AUR. Supported floors are Windows 10 and 11 on x64 and arm64, macOS 12 and newer on Intel and Apple Silicon, and Ubuntu 20.04 and newer, Fedora 36 and newer, Arch and similar distributions. One macOS wrinkle: the app isn’t yet Apple-signed, so a damaged-app warning needs sudo xattr minus c and r on the bundle before first launch. GPUI Help Check Updates uses signed manifests for supported install types, while Homebrew and deb or rpm installs update through their own package managers.
NyaTerm Against the Usual Suspects
Item
NyaTerm
MobaXterm
Tabby
WindTerm
PuTTY
SecureCRT
License and source
MIT, fully open
Proprietary, paid Pro tier
MIT, open with plugins
Free of charge, source quiet
MIT, open source
Proprietary, paid
Interface stack
Rust plus GPUI native, GPU rendered
Native Windows
Electron browser engine
Native, framework closed
Native Win32
Native
SSH, Telnet, serial, SFTP
Yes, all four plus tunnels
Yes, plus X server
SSH, Telnet, serial, SFTP
Yes, all four
SSH, Telnet, serial, SCP via tools
Yes, all plus scripting
RDP and VNC panes
Yes, isolated helper processes
Yes, embedded
Via plugins and profiles
Partial
No
Yes
Session sync and backup
Encrypted WebDAV, S3, Gist plus .nya files
Settings export, cloud limited
Cloud vault plus local config
Local profiles
Registry or files
Cloud plus local
OTP and credential autofill
Built in TOTP and HOTP
Password vault
SSH vault plus MFA notes
Basic
No
Credential vault
AI assistance
Built in Ask and Agent modes
None
AI plugins
None
None
None
Resource weight
Light native process
Medium
Heavy Electron
Light
Featherweight
Medium
Best for
Open source all-in-one remote workspace
Windows admins wanting everything bundled
Plugin tinkerers who accept Electron
Speed fans comfortable with stalled upstream
Single quick connections
Enterprises buying support contracts
The honest read: PuTTY still wins the five-second

https://hamradio.my/nyaterm-remote-terminal-workspace-ssh-sftp/

#remoteterminal #terminalworkspace #ssh #terminalclient #sftp #serial #rdp #nyaterm #windows #telnet

#rss #feeds #linksharing #qwibbit

Disk Cloning for the Shack: Backing Up Your PCs from Norton Ghost ...

2026-11-03T15:19:31Z

Disk Cloning for the Shack: Backing Up Your PCs from Norton Ghost to Clonezilla

TL;DR: Disk cloning Clonezilla style is how a ham shack survives a dead SSD or a corrupted update: copy the whole drive, boot sectors and all, so a restore takes an hour instead of a rebuild weekend. Norton Ghost ruled this world in the 2000s and was discontinued in 2013. Its open source successor is Clonezilla, free and still actively developed, with Rescuezilla as its friendly face, FOG Project for cloning whole club fleets over the network, and rpi-clone for Raspberry Pi SD cards. Image your machines now, label the images, test the restore, and keep a rescue USB in the go kit.
What Disk Cloning Is
Disk cloning is the sector-by-sector copying of an entire drive, partitions, boot loader, operating system, drivers, applications, settings, and data, either into a compressed image file for later restore or straight onto another drive for immediate swap-in. Three terms get mixed up constantly, so here is the short version:
Method
What it copies
Restores a bootable PC?
File backup
Selected files and folders
No, needs a working OS first
Disk image
Every used sector into one file
Yes, via a rescue USB
Disk clone
Every used sector straight onto another drive
Yes, swap the drive and boot
Good tools copy only used sectors by understanding the filesystem, so a 1 TB drive holding 60 GB of data produces a roughly 30 to 60 GB compressed image instead of a 1 TB one. Bad or ancient tools copy every sector blindly, slow and wasteful. After writing, always verify the image. An unverified backup is a rumor, not a backup.
Rescuezilla welcome screen with Backup, Restore, Clone, Verify Image and Image Explorer. Rescuezilla reads and writes Clonezilla compatible images with a point and click GUI. Screenshots: Rescuezilla project, GPL.
Why Hams Should Clone Their Disks
File backups save your logs and configs. They don’t save you when the shack PC refuses to boot at 11 pm before a contest, when a Windows update bricks the digital modes machine, or when the SD card in the APRS igate dies silently after three years of continuous logging. A disk image is a snapshot of everything: partitions, boot loader, operating system, drivers, applications, settings, and data. Restore it to a fresh drive and the machine comes back exactly as it was.
Three jobs cover nearly every amateur need:
Backup – take an image of a working machine and store it on an external drive. When disaster strikes, restore.
Duplicate – copy one known good setup to more hardware. One configured DietPi APRS igate becomes five identical igates. One field day logging laptop becomes six.
Migrate – move everything to a bigger SSD without reinstalling anything.
Cloning is the unglamorous insurance policy behind every reliable shack, club station, repeater site, and emergency communications deployment.
Norton Ghost: The King of the 2000s
If you worked in IT between roughly 1998 and 2010, you knew Ghost. Understanding its story explains why the whole field works the way it does today.
Origins in Auckland
GHOST, an acronym for General Hardware-Oriented System Transfer, was written in 1995 by Murray Haszard for a small New Zealand company called Binary Research in Auckland. It was a DOS program that could copy a hard drive sector by sector and squeeze the result into .gho image files, and it was remarkably fast for its era. Symantec bought the technology in 1998, and development stayed in Auckland until 2009. The full product history is on Wikipedia’s Norton Ghost page.
The golden age: Ghost 2003
The version everyone remembers is Norton Ghost 2003, released in September 2002 and bundled with Norton SystemWorks 2003 Professional. You made a DOS boot floppy, booted the dead PC from it, and pulled an image back over the network. It just worked, on hardware its own installer barely understood, and an entire generation of technicians learned their trade from it.
Ghost’s killer feature for institutions was GhostCast Server: multicast one image to a whole classroom, computer lab, or office floor simultaneously. Schools and universities imaged hundreds of identical PCs in an afternoon. At its peak Ghost was the default deployment tool of corporate IT, bundled in the yellow Norton boxes sold in every computer shop. Its popularity had a shadow side too: the same .gho convenience made it the standard format for pirated pre-activated Windows distributions across Asia in the 2000s.
The PowerQuest takeover and the long decline
At the end of 2003 Symantec bought its biggest competitor, PowerQuest, and in August 2004 released Norton Ghost 9.0 built on PowerQuest’s Drive Image 7 codebase instead of the original Ghost engine. The image format changed to .v2i and was incompatible with old .gho files. Old hands considered it a different product wearing a familiar name, and the classic DOS based engine survived only in the enterprise Ghost Solution Suite line.
Versions 10 through 15 followed, the last in 2009 to 2010, but the world had moved on: UEFI firmware, GPT partition tables, BitLocker encryption, and Linux filesystems like ext4, which only arrived in the enterprise edition in 2017, long after the consumer line was dead. Symantec discontinued Norton Ghost on April 30, 2013, with support ending June 30, 2014, and pointed customers at Symantec System Recovery. The enterprise Ghost Solution Suite lingered on under Broadcom ownership, but for ordinary users the Ghost era was over.
Ghost’s legacy is the vocabulary itself. Hams of a certain age still say “ghost the drive” the way people say “xerox a copy”. But nobody should be ghosting anything in 2026: the last consumer release can’t handle UEFI boot, modern NVMe drives, or current filesystems, and it has received no fixes in over a decade.
Clonezilla: The Open Source Successor
While Ghost was declining, its replacement was already growing in Taiwan. Clonezilla was designed by Steven Shiau and developed by the NCHC Free Software Labs (National Center for High-Performance Computing), with the project starting around 2003 to 2004. It is free software under the GPL, and unlike abandonware it’s still actively developed, with releases rolling through 2026 on modern kernels. Downloads live at clonezilla.org/downloads.php.
Clonezilla comes in two flavors:
Clonezilla Live – boots from USB, clones one machine to an image or straight to another disk. This is the one hams want.
Clonezilla Server Edition (SE) – deploys one image to many machines over the network, the open source answer to GhostCast.
Under the hood it is a smart integration of best of breed open source parts: Partclone (by Thomas Tsai, also NCHC) for filesystem aware copying, Partimage, ntfsclone, and udpcast for multicast. It understands ext4, NTFS, FAT, HFS+, and current releases have restored btrfs support via Partclone 0.3.36. Recognition followed: SourceForge Project of the Month in January 2010, millions of downloads, and a permanent place in every technician’s USB toolkit.
Why it beat Ghost for our purposes: it costs nothing, it handles Linux filesystems natively, it boots on UEFI and NVMe without drama, and the multicast server edition is free for clubs that need to image a fleet.
Modern Cloning Tools Compared
Tool
Cost
Interface
Best for
Clonezilla Live
Free, open source
Text menus
The default choice for whole disk backup and restore
Rescuezilla
Free, open source
Graphical
Beginners, anyone scared of Clonezilla menus, fully Clonezilla image compatible
FOG Project
Free, open source
Web console
Clubs imaging many PCs over the network, PXE boot fleets
Redo Rescue
Free, open source
Graphical
Simple point and click rescue imaging
Foxclone
Free
Graphical
Linux Mint and Ubuntu desktops
Veeam Agent
Free edition
Windows GUI
Scheduled image backups of a running Windows shack PC
Hasleo Backup Suite
Free
Windows GUI
Modern free Windows imaging after Macrium’s exit
Macrium Reflect
Paid subscription
Windows GUI
Polished Windows tool, but the free edition was retired in January 2024 and Reflect X is subscription only
Acronis / vendor tools
Paid or bundled
Windows GUI
Samsung Data Migration and similar are fine for one way SSD upgrades
dd / ddrescue, GNOME Disks
Free, built in
Command line / GUI
Raw copies, damaged drives (ddrescue), quick SD card dumps
Two notes for the nostalgic. First, old Macrium Reflect Free v8 still floats around download sites, but it stopped receiving even security patches in January 2024, so don’t build a backup strategy on it. Second, anything that only runs on DOS era boot media belongs in a museum next to the floppy disks, not in your go kit.
Workflows for the Shack
Workflow 1: Back up a Windows shack PC
Download Rescuezilla, flash it to a USB stick with Balena Etcher or Rufus.
If the PC uses BitLocker, suspend protection first so the rescue environment can read the drive.
Boot the USB, choose Backup, select the internal drive, and save the image to an external USB drive.
Let it verify, then label the image with machine, date, and Windows version, for example shack-hp-EliteDesk_win11_2026-11-03.
Once a year, actually restore to a spare drive and confirm it boots. An untested image is a hope, not a plan.
Workflow 2: Duplicate a Raspberry Pi node
This is the bread and butter for APRS igates, DX cluster nodes, and ADS-B feeders:
While the Pi is still running and healthy, attach a second SD card over a USB reader and run rpi-clone. It copies the live system to the spare card.
Alternatively power the Pi down, pull the card, and take an image with Raspberry Pi Imager or dd on another machine.
Boot the clone once to confirm it works, then bag it with a label and keep it at the site or in the go kit.
Tip: give each duplicate its own hostname and, for network services, check for address or callsign collisions before putting two copies on the same LAN.
Workflow 3: Clone a Linux SSD to a bigger one
Connect the new SSD over USB, boot Clonezilla Live.
Use device-device mode, disk_to_local_disk, source old drive to target new drive.
Let Clonezilla grow the partitions with the -k1 option, then boot the new drive and expand the filesystem to fill it.
Special case that bites people: a btrfs multi device pool spread over two partitions or drives is one filesystem wearing two hats. Clone the entire disk, never a single partition, or the pool won’t reassemble. Current Clonezilla releases understand btrfs, but whole disk mode is still the safe path.
Workflow 4: Golden rules for all of the above
Stop the databases first. A live MariaDB, DXSpider spot database, or logging program copied mid write restores corrupted. Shut down containers or services, or use filesystem snapshots, before imaging.
Verify every image and test boot the important ones.
Label everything: machine, OS, date. Future you won’t remember what backup-final-2.img contains.
Keep the rescue USB with the images. An image without bootable restore media is a paperweight.
Cloning Windows to different hardware may need driver repair and reactivation. For fleets of identical club PCs this rarely matters; for mixed hardware, look into sysprep generalize before capturing the master image.
Retire spinning rust images regularly. Re-image after major changes: new radio software, antenna rotator drivers, contest logger upgrades.
Gotchas Worth Knowing
UEFI vs legacy boot: modern PCs boot UEFI from GPT disks. Clonezilla and Rescuezilla handle both, but make sure the rescue USB itself boots in UEFI mode or the restored disk may not start.
Secure Boot: current Clonezilla Live releases boot fine with Secure Boot on. Ancient tools don’t.
Smaller target drives: images restore to equal or larger drives. Shrinking partitions to fit a smaller SSD is possible but fiddly; avoid it by buying the bigger drive.
4K sector drives: mixing old 512 byte and new 4K native sector drives can misalign partitions and cost performance. Whole disk clone tools normally handle alignment, partition only copies sometimes don’t.
SD card wear: cards in 24/7 Pi service die of write exhaustion. Imaging the card preserves the software, but replace aging cards proactively and keep writes low with tmpfs and RAM logging.
Conclusion
Norton Ghost earned its legend: a fast DOS tool from Auckland that taught the world what disk imaging could be, then faded when its owner stopped keeping up with UEFI, modern filesystems, and Linux. Clonezilla picked up the torch as free software and never put it down. For a ham, the practical stack is small: Rescuezilla or Clonezilla Live on a USB stick for the PCs, rpi-clone for the Pi fleet, FOG Project if the club runs a training room, and a labelled external drive holding tested images. Pair the backup habit with disk encryption on portable drives, covered in the VeraCrypt guide, and build the images on a quiet Sunday. Then the next dead drive becomes a story about a one hour restore instead of a lost contest weekend.
Sources and Further Reading
Clonezilla official site and downloads: https://clonezilla.org/
Clonezilla Live download page: https://clonezilla.org/downloads.php
Rescuezilla official site and download: https://rescuezilla.com/
FOG Project community imaging platform: https://fogproject.org/
Partclone filesystem aware imaging library: https://partclone.org/
Rpi-clone live Pi cloning script: https://github.com/billw2/rpi-clone
Redo Rescue point and click rescue imaging: https://github.com/redorescue/redorescue
Norton Ghost product history: https://en.wikipedia.org/wiki/Norton_Ghost
Encrypt portable backup drives with VeraCrypt: https://hamradio.my/veracrypt-disk-encryption-ham-radio/
73 de 9M2PJU
The post Disk Cloning for the Shack: Backing Up Your PCs from Norton Ghost to Clonezilla appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/disk-cloning-shack-pc-backup-norton-ghost-clonezilla-guide/

#nortonghost #diskcloning #clonezilla #shack #backing #restore #backup #every #usbdisk #rescuezilla

#rss #feeds #linksharing #qwibbit

DeepCW: Ultra-Accurate Neural Network Morse Code Decoder That ...

2026-11-03T11:20:25Z

DeepCW: Ultra-Accurate Neural Network Morse Code Decoder That Runs in Your Browser

What DeepCW Is
DeepCW is a free real-time Morse code decoder powered by a neural network model that runs entirely in the web browser with no install and no server side audio upload.
If you have ever watched a classic CW decoder fall apart the moment a signal fades into the noise, drifts slightly off frequency, or gets stepped on by a stronger station 200 Hz away, you already know the problem DeepCW was built to solve. Traditional decoders use fixed tone detectors, timing thresholds and character templates. They work well on loud clean machine sent practice files and collapse on real HF. DeepCW replaces that fragile pipeline with a trained model that listens the way an experienced operator listens, by recognizing the shape of Morse through fading, interference and noise. You open https://deepcw.cc/, allow microphone or route radio audio, and readable text starts flowing. The source lives at https://github.com/e04/web-deep-cw-decoder under the MIT license, with 106 stars and 20 forks at the time of writing.
DeepCW running live at deepcw.cc with the decoder ACTIVE, spectrum traces visible and single channel 800 Hz mode selected. Screenshot by 9M2PJU.
This guide covers why classic decoders struggle, how DeepCW works under the hood, what the published benchmark numbers actually mean, how it compares against CW Skimmer, fldigi and ggmorse on real on air clips, how to wire it into a real shack, and where its honest limits sit.
Why Classic CW Decoders Hit a Wall
Every CW decoder faces the same four enemies, and fixed DSP logic handles each one poorly.
First is weak signal work. A detector that triggers on energy above a threshold has a brutal choice. Set the threshold low and noise prints as garbage characters. Set it high and weak but perfectly readable Morse vanishes. Human ears integrate over time and track rhythm. Thresholds can’t.
Second is fading and QSB. On HF a signal can swing 10 dB or more inside a single transmission as ionospheric paths add and cancel. A detector tuned for the peaks misses the troughs. AGC helps audio comfort but smears the on off timing that decoders depend on.
Third is speed and spacing variation. Real operators, especially in contests and pileups, run together words, clip spaces, stretch dahs, and send with paddles, bugs and straight keys that all sound different. Template matching expects textbook timing. Humans adapt in seconds.
Fourth is crowded band conditions. A loud station nearby, key clicks, power line buzz, static crashes and overlapping callers all land inside a wide receiver passband. A single tone detector with no context can’t tell which energy belongs to the target.
The result is familiar to anyone who has tried decode software on 40 meters at night. It copies the loud CQ caller fine, prints nonsense during the weak reply, and gives up entirely when two stations overlap. DeepCW attacks this at the model level instead of adding more knobs to the same detector.
How DeepCW Works Under the Hood
DeepCW ships as a modern web app built with Vite, with the decode model running locally in the browser. There’s no audio upload to a cloud server, which keeps latency low and keeps operating viable on modest links once the page has loaded. The project advertises cross platform support for Windows, macOS, Android and iOS, which in practice means any modern Chromium or WebKit based browser that can grant microphone access and run WebAssembly or ONNX inference.
The heart of the system is a trained neural network distributed as ONNX files in the companion repository deepcw-engine. Two models are provided. The full model.onnx handles general decoding, while the faster narrow band model_narrow.onnx is intended for multi-channel decoding where several instances run in parallel. Metadata in model.onnx.json defines the expected input, currently mono 16 bit PCM resampled to 3200 Hz. The documented preparation step makes this explicit:
ffmpeg -i input.wav -ac 1 -ar 3200 -sample_fmt s16 target.wav
If the metadata ever changes, the sample_rate field in that JSON file is the authority, not the example above.
The Python and Node.js examples in deepcw-engine show how thin the wrapper is. Install dependencies, point the script at the model, the metadata and a WAV file, and text comes out:
cd examples/python
python3 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt
python decode_morse.py \
--model ../../model.onnx \
--metadata ../../model.onnx.json \
--wav test.wav
cd examples/nodejs
npm install
npm run decode -- \
--model ../../model.onnx \
--metadata ../../model.onnx.json \
--wav test.wav
That thin wrapper matters for hams who build tools. The browser app is the friendly face, but the engine is a reusable Morse recognition block that can be embedded in loggers, skimmers, monitoring nodes and accessibility aids.
The live UI exposes exactly what an operator needs and nothing more. DECODER toggles between OFF and ACTIVE. MODE selects SINGLE or multi-channel operation. WINDOW sets the visible spectrum history, 6 seconds in the screenshot above. WIDTH sets the decode bandwidth around 800 Hz. INPUT selects the audio source. THRU controls audio pass-through. A settings gear opens deeper options including RTTY mode. Everything fits on a phone screen, which is deliberate, because many operators run it on a tablet beside the rig.
Benchmark Numbers That Actually Mean Something
The README publishes a character error rate heatmap measured in HIGH mode, and the methodology is stated clearly enough to trust. Signal power is time averaged over the whole record at roughly 50 percent keying duty cycle, referenced to a 2500 Hz noise bandwidth in additive white Gaussian noise. Error rate is CER, the percentage of inserted, deleted or substituted characters relative to the reference text.
The headline results are striking:
SNR in 2500 Hz bandwidth
Character error rate
What that feels like
0 dB to -4 dB
0.00 percent at all tested speeds
Perfect copy where a human operator is still comfortable
-6 dB
Nearly error free
Occasional doubt on a character, solid QSO copy
-8 dB
Below 1.5 percent
Usable weak signal copy, a wrong character every few words
-10 dB
Below 8 percent
Hard copy, callsigns need repeats but meaning survives
For context, -10 dB in a 2500 Hz bandwidth is deep into territory where classic decoders print confetti. A skilled human with narrow filters can still pull callsigns through it. That DeepCW stays under 8 percent errors there, and perfect above -4 dB, puts it in the same conversation as a focused operator rather than a toy decoder.
Two caveats keep this honest. AWGN testing is the clean laboratory case. Real HF adds selective fading, interference, chirp, key clicks and overlapping signals that are crueler than white noise. And CER counts characters, not QSO outcomes. A 2 percent CER that eats the two letters that distinguish similar callsigns hurts more than a 2 percent CER scattered across ragchew prose. Still, as a controlled comparison baseline, this is far more transparent than the anecdotal “copies great” claims most decoder pages offer.
DeepCW Versus CW Skimmer, fldigi and ggmorse
To ground the claims, the author compared DeepCW against three established tools on public short CW QSO videos from YouTube: CW Skimmer, fldigi and ggmorse. All tested software was the latest available version as of 3 June 2026. The author is explicit that this isn’t a general ranking, only a comparison on the specific clips, settings and evaluation method shown.
Video 1 is a fast contest style exchange with heavy repetition and short prosigns. The reference reads AI5DD AI5DD 56N CO BK BK GA UR 55N 55N OK OK 73 AE0Q DE AI5DD 44 EE R 44 EE EE. DeepCW holds the structure together with only spacing slips, 55N55N merged and a fused 73AE0QDE, but every callsign and exchange element survives. CW Skimmer invents a plausible but wrong opening UI5DDM EU and later hallucinates MTKE. Fldigi fragments badly with single letter shards. Ggmorse keeps the callsign but drops exchange markers like CO into noise.
Video 2 is a clean JA to US style QSO, reference D DE JO2QOT JO2QOT 5NN CA 5NN 100 TU JO2QOT TU K6XX. DeepCW reads D DE J02Q O T J02QOT 5NN CA 5 NN 100 TU J02QOT TU K6XX, confusing the letter O with zero twice but otherwise delivering a loggable QSO. CW Skimmer splits the exchange across lines and drops the opening. Fldigi inserts stray diacritics and drops a callsign instance. Ggmorse holds up reasonably but garbles the report context with SEGE.
Video 3 is the cruel one, a long ragchew style exchange with signal reports, BK breaks and sign offs, reference starting ? WD4DAN WD4DAN GE ES FB UR 57N 57N CO BK BK TU GE UR 56N 56N GA GA 73 BK BK TU GA 73 DE W0ABE TU EE. Here every decoder degrades, which is exactly the point. DeepCW keeps the skeleton readable with merged words like ESFB and BKTUGA but preserves both callsigns and the report flow. CW Skimmer interleaves two overlapping decode lines, fldigi collapses into punctuation soup, and ggmorse produces long error runs of repeated E and T characters.
The pattern across all three clips is consistent. DeepCW isn’t perfect, and nobody should log directly from any decoder without ear verification. But where classic tools insert confident garbage or shatter words into fragments, DeepCW degrades like a tired operator, merging spaces and confusing similar characters while keeping the QSO structurally intact. That’s the signature of a sequence model with learned language context rather than a symbol by symbol detector.
Decoder
Strong clean copy
Weak and faded copy
Overlapping signals
Readability of errors
DeepCW
Excellent, spacing slips only
Usable to -8 dB, graceful degradation
Best of the four tested, holds structure
High, errors look like tired copy
CW Skimmer
Excellent, built for contest skimming
Good on loud callers, invents text when weak
Good multi decoder design, splits lines
Medium, confident wrong calls hurt
fldigi
Good with careful tuning
Fragile, fragments into shards
Poor without manual filter help
Low, punctuation soup
ggmorse
Good on clean machine sent CW
Mixed, long E and T error runs
Fair
Medium, error bursts obvious
CW Skimmer remains the king of wideband contest skimming across many kilohertz with cluster integration. Fldigi remains the Swiss army knife for dozens of modes beyond CW. Ggmorse remains a neat lightweight option for clean signals. DeepCW is the one to reach for when the signal is weak, fading or crowded and you need one stream copied well.
Wiring DeepCW Into a Real Shack
Setup takes minutes because there’s nothing to install.
Open https://deepcw.cc/ in a modern browser on the computer or tablet that will decode. Press the DECODER tile so it flips from OFF to ACTIVE. Grant microphone permission when asked. For a first test, hold a speaker near the microphone or play one of the sample clips and watch text appear in the lower pane while signal traces draw across the spectrum.
For real operation, never rely on acoustic coupling. Route audio with a cable:
From a rig with USB audio, select the rig sound device as INPUT in DeepCW or as the system default input before opening the page.
From a rig with line out or headphone out, use a USB sound dongle plus an isolation transformer to avoid ground loops, then select that device.
On Windows, use VB-Cable or Virtual Audio Cable to pipe SDR software into the browser. On Linux, use PipeWire loopbacks or PulseAudio null sinks. On macOS, use BlackHole.
Keep the receiver bandwidth around 300 to 800 Hz for single channel work and center the target near the WIDTH passband shown in the UI.
Practical settings that work for most HF use:
MODE SINGLE for one QSO, multi-channel only when monitoring a pileup or a contest band edge.
WINDOW 6 seconds to see enough context for rhythm without clutter.
WIDTH 800 Hz to tolerate drift, narrower once locked on a stable caller.
THRU OFF for silent decoding beside the rig speaker, ON when you want the neural noise reduced audio in your headphones.
RTTY mode from settings when switching to the DeepRTTY engine on a RTTY signal.
Three habits separate good copy from frustration. First, give the model a few seconds of signal before judging it, since sequence context builds over the WINDOW. Second, fix audio level so the target trace is clearly visible but not clipping, because crushed audio destroys timing. Third, verify callsigns by ear before logging. Decoders confuse O with zero, S with H under fading, and merge word spaces when operators run words together. Use DeepCW as a second operator who never tires, not as a legal logger.
Mobile operation is genuinely usable. A tablet running DeepCW beside an IC-705 or a KX2 on a summit decodes while you operate, and the large ACTIVE and SINGLE tiles are glove friendly. Pack a short USB-C audio interface rather than fighting phone microphone pickup in wind.
Use DeepCW on cw.hamradio.my
9M2PJU hosts the same open source DeepCW web app at https://cw.hamradio.my/, so Malaysian and Southeast Asian operators have a nearby copy alongside the upstream site at deepcw.cc. Either address runs the identical browser based decoder. All inference happens locally in the page, so your radio audio never leaves the machine beyond the page you opened. If one address is slow or unreachable on your network, try the other before touching any settings, since most first run failures are network or audio routing issues rather than decoder problems. Bookmark the 9M2PJU copy for field use and keep the upstream address as backup, or the reverse if the upstream loads faster from your location.
Multi-Channel Decoding and Neural Noise Reduction
Two features lift DeepCW above a single stream decoder.
Multi-channel mode uses the faster narrow band model to track several CW signals at once. Contest operators know this pain well. The run frequency plus two nearby callers plus a DX station 400 Hz up, all relevant, all overlapping in a wide waterfall. Classic single tone decoders force a choice. Multi-channel mode lets each slice get its own model instance. CPU cost rises, so a modest laptop or a recent phone handles a few channels while an old tablet should stay in SINGLE mode.
Audio pass-through with neural noise reduction is the quieter revolution. Instead of only printing text, DeepCW can return cleaned audio through THRU so noisy CW is easier on the ears during a long session. This work connects directly to the sibling project HamNoise, an entirely local neural noise reduction system for CW and voice including SSB that runs as a web app at https://e04.github.io/HamNoise/. The comparison page frames it well. RNNoise runs realtime but targets generic voice, NVIDIA RE-USE sounds excellent but processes whole files offline, and RM Noise needs a server round trip. HamNoise runs locally in realtime and is tuned for ham audio. Long CW sessions with static crashes and band hiss become far less fatiguing, and weak Morse pops out of the muck in a way that narrow analog filters alone never achieve.
DeepRTTY: The Same Brain on a Different Mode
RTTY operators get the same treatment. DeepRTTY applies the same deep learning approach to radioteletype and is selectable inside DeepCW through the settings RTTY mode. The README pitch is direct. It recovers copy where conventional demodulators fail under weak signals, QSB and heavy noise. Anyone who has watched classic RTTY software print gibberish as the diddle fades knows how valuable that is, especially on crowded contest weekends when signals sit shoulder to shoulder. If DeepCW proves itself on your CW workflow, switching the same page to RTTY mode for a sprint is a zero install bonus.
For Builders: deepcw-engine, HamNoise and Your Own Tools
The open architecture is what makes this project interesting beyond one web page. The deepcw-engine repo holds the ONNX models, metadata and minimal Python and Node.js examples under MIT. A builder can decode WAV archives of contests, wire recognition into a KiwiSDR or OpenWebRX extension, add Morse alerts to a monitoring node, or prototype accessibility tools that turn beacon IDs into text.
A sensible builder path looks like this. Start with the ffmpeg normalization to mono 3200 Hz 16 bit PCM. Run the Python example on a known recording and measure CER against your own transcript rather than trusting the headline numbers on yo

https://hamradio.my/deepcw-neural-network-cw-decoder-browser/

#morsecode #neuralnetwork #codedecoder #deepcw #browser #runs #real #classic #wallevery

#rss #feeds #linksharing #qwibbit

Ambien: An ESP32 CO2 Sensor With E-Paper That Runs a Year on Two ...

2026-11-03T10:00:56Z

Ambien: An ESP32 CO2 Sensor With E-Paper That Runs a Year on Two 18650s

TL;DR: Ambien is an open source ESP32 CO2, temperature and humidity sensor with a 4.2 inch e-paper display that runs up to a year on two 18650 cells. It publishes one JSON message over MQTT with Home Assistant auto discovery, works fully offline, and stays silent for RF quiet shacks. Hardware tour, battery math and shack workflows inside.
What Ambien Is
Ambien is an open source ESP32 based CO2, temperature and humidity sensor with a 4.2 inch e-paper display that wakes, measures, refreshes the screen and returns to deep sleep, stretching two 18650 cells to about a year of operation.
It is the work of mcer12, published under GPL-3.0 and declared done, with firmware, EasyEDA PCB sources and gerbers, 3D printable enclosure files and even unit tests in one repo. A Sensirion SCD4x measures CO2 while an SHT30 covers temperature and humidity. Every wakeup publishes one JSON message over MQTT with native Home Assistant discovery, and a mobile friendly captive portal handles setup with a theme switch. Skip Wi-Fi entirely and it still works as a silent offline display. The pixel style interface uses the author PixelFonts project for a retro paper look that suits e-paper perfectly.
Ambien on the shelf doing its one job. Photo: mcer12 project repo, GPL-3.0.
Why a Shelf Sensor Belongs in a Ham Shack
Shacks are small rooms full of warm bodies, hot radios and closed doors. A two person contest station overnight easily pushes CO2 past 1,000 ppm, the point where drowsiness and dull thinking set in, exactly when you need sharp ears for weak signals. Most operators blame fatigue. Some of it is chemistry, and a 50 dollar sensor answers the question objectively.
E-paper brings a second shack advantage that rarely gets mentioned. Unlike TFTs with PWM backlights and buzzing boost converters, an e-paper display draws nothing at rest and emits nothing while you read it. No switching noise near HF receivers, no glow wrecking night vision during meteor scatter sessions, no fan, no hum. It's the most RF polite display technology available, and this project exploits that fully.
The third reason is architectural. Ambien is a masterclass in duty cycling an ESP32, the same chip family running fox hunt beacons, DX cluster clients and LoRa trackers across this site. Its sleep discipline, Wi-Fi on for seconds, sensors in low power modes, everything else dark, is directly reusable in field nodes that must survive weekends on small cells.
Hardware Tour: Overkill MCU, Frugal Everything Else
ESP32 Because Spares Existed
The author admits the ESP32 is massive overkill and used one because spares sat in the drawer. Honest engineering beats optimal engineering when the alternative is a three week parts order. The chip spends nearly its whole life in deep sleep anyway, so the overkill costs nothing in battery terms and buys USB native upload, ample RAM for the display buffer and Wi-Fi when needed.
E-Paper Display, 4.2 Inch, 400 by 300
The screen is a 4.2 inch 400 by 300 black and white e-paper panel, HINK-E042A13-A0 or equivalents with the centered flex cable. That resolution fits big readable digits across a room, and the bistable nature means the last reading stays visible for months with zero power. Power fails during a storm and the shack whiteboard still shows the last CO2 value. Try that with an LCD.
Sensirion SCD4x Plus SHT30
CO2 comes from the Sensirion SCD4x photoacoustic sensor family, temperature and humidity from the SHT30. Both are the same parts serious commercial monitors use, not the drift prone eCO2 estimation chips that guess CO2 from VOCs. The firmware offers forced recalibration to 430 ppm from the web interface, which you trigger with the unit breathing fresh outdoor air. The CO2 sensor is optional in software. Disable it and the layout switches to a big temperature plus humidity face while battery life multiplies.
Power: Two 18650s, TP4057, Optional Protection
Two 18650 slots plus a TP4057 charger and USB-C cover the power story, with an optional FS312F-G protection chip at a higher cutoff to protect cell health. Prefer a flat pack? Omit the slots and protection and wire a LiPo instead. The headline numbers fall straight out of the duty cycle. About a year with CO2 sensing, multiple years on temperature and humidity alone.
Pixel font readings using the author PixelFonts project. Photo: mcer12 project repo.
Battery Math Worth Stealing
Two quality 18650s hold roughly 7,000 mAh combined. Spread over 8,760 hours in a year, the average budget is about 800 microamps. The ESP32 in deep sleep sips single digit microamps, the e-paper needs power only during its few second refresh, and Wi-Fi transmits for seconds per wakeup. The SCD4x in low power mode dominates the budget, which is exactly why disabling CO2 multiplies life into multiple years.
Any builder of battery field nodes should memorize this shape. Capacity divided by hours gives the average current, and every subsystem must fit under it. Ambien publishes the answer in working hardware rather than spreadsheet theory.
Software Tour: One JSON Message and a Portal
MQTT in a Single Publish
Each wakeup sends one non retained message to <topic>/state, default topic ambien/ambien_xxxx, shaped like this:
{"temperature":23.5,"humidity":45,"co2":812,"battery":4.05}
One message, four fields, no chatty protocol. Battery voltage rides along every time, so Home Assistant can graph cell decline across seasons and warn before winter kills a weak cell.
Home Assistant Discovery With One Checkbox
Tick Register in Home Assistant and the next wakeup publishes a retained discovery config to homeassistant/device/ambien_xxxx/config. Entities appear automatically with the device grouped correctly. The checkbox then unticks itself, a small UX touch that prevents duplicate registrations. Readers building the rest of the smart shack should start with the companion Home Assistant hardware and installation guide.
Captive Portal and Offline Mode
First boot with no Wi-Fi starts an Ambien_XXXX access point serving a mobile friendly setup page at 192.168.0.1, with light and dark themes. Enter Wi-Fi, wake interval and MQTT details, save and reboot. Later changes come from the AP MODE or LOCAL SERVER buttons on the device. The OFFLINE MODE button on the first run screen skips networking entirely for a pure display that never transmits, ideal for exam halls, bedrooms and anyone who simply dislikes another Wi-Fi client.
Setup portal with theme switch. Photo: mcer12 project repo.
Build and Setup Path
The project uses PlatformIO with libraries fetched automatically, so firmware is one command:
pio run -t upload
MQTT payload unit tests run on the PC with pio test -e native, or on hardware with pio test -e esp32dev connected. PCB sources live in EasyEDA format with gerbers included, and the enclosure ships as three 3MF files for top, bottom and holder. Setup follows the access point flow above, then one forced 430 ppm calibration outdoors before trusting absolute CO2 numbers.
Ambien Versus Commercial CO2 Monitors
Item
Ambien
Typical e-ink commercial monitor
Always on Wi-Fi monitor
Sensors
SCD4x true CO2 plus SHT30 temp and humidity
True CO2 plus temp and humidity
True CO2 plus temp, humidity, sometimes VOC and PM
Display
4.2 inch e-paper, room readable
Small e-ink, room readable
Small color LCD, needs power
Battery
2×18650, about a year with CO2
Coin or AA cells, one to two years
USB powered, no battery operation
Smart home
MQTT plus Home Assistant discovery
Vendor app and cloud, sometimes Matter
Vendor app plus Matter or HA
Offline use
Full display only mode
Full display only mode
Dead without power and Wi-Fi
Data path
Your MQTT broker, nothing leaves home
Vendor cloud account usually required
Vendor cloud account usually required
Openness
GPL-3.0, PCB, enclosure, tests
Closed
Mostly closed
Cost shape
Parts plus an evening
Mid range retail
Budget to mid range retail
Best for
Hams and makers who own their data
Bedrooms and classrooms wanting zero setup
Desks already full of USB chargers
Shack Workflows That Earn the Shelf Space
Park one at eye level behind the operating position. Glance up between overs. CO2 climbing toward four digits means open the door or crack a window before the next hour, not after the headache. Log weekend contest CO2 alongside scores for a season and the ventilation lesson writes itself.
Pair readings with automation. Home Assistant can boost a USB fan, flash the shack lamp or send a phone alert when CO2 crosses your threshold. Battery voltage in every message means the sensor joins the same low battery watch as your handhelds.
Take it portable. Field Day tents, exam sessions and club meetings all concentrate people in still air. Offline mode keeps it silent and emission free, and the year long battery means it lives in the go box without a charging rota.
Honest Limits
Sixty five stars means a young community, so expect to answer your own questions from the source. The e-paper refreshes in seconds, useless for real time graphs. The 4.2 inch panel and dual 18650s make a chunky shelf object, not a pocket meter. Forced calibration demands genuine fresh air, and calibrating beside a running generator invents accuracy problems. The project is declared done rather than actively developed, which is a feature for stability and a risk if Espressif breaks an API under it.
Sources and Further Reading
Ambien project repo, firmware, PCB and enclosure: https://github.com/mcer12/Ambien-ESP32
PCB sources, schematic and gerbers: https://github.com/mcer12/Ambien-ESP32/tree/main/pcb
3D printable enclosure models: https://github.com/mcer12/Ambien-ESP32/tree/main/enclosure
PixelFonts retro font project: https://github.com/mcer12/PixelFonts
Home Assistant MQTT integration docs: https://www.home-assistant.io/integrations/mqtt/
Home Assistant hardware and installation guide: https://hamradio.my/home-assistant-best-hardware-installation-guide/
73 de 9M2PJU
The post Ambien: An ESP32 CO2 Sensor With E-Paper That Runs a Year on Two 18650s appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/ambien-esp32-co2-epaper-sensor/

#ambien #two #runs #homeassistant #shack #sensorbelongs #json #shelfsensor #overkill #everythingelse

#rss #feeds #linksharing #qwibbit

OWLOS Flight Radar: Live ADS-B Air Traffic on a $15 ESP32 Display ...

2026-11-03T09:21:12Z

OWLOS Flight Radar: Live ADS-B Air Traffic on a $15 ESP32 Display

TL;DR: OWLOS Flight Radar turns 10 to 30 dollar ESP32 display boards into live ADS-B scopes fed by free community networks, flashed from the browser and driven by touch. Four boards supported, scriptable USB console, maps cached for offline viewing. Board guide, scope reading and ham feeder tie-ins inside.
What OWLOS Flight Radar Is
OWLOS Flight Radar is an open source ESP-IDF firmware that turns cheap ESP32 display boards into live air traffic scopes, drawing nearby aircraft as heading turned silhouettes colored by altitude with trails behind them.
Tap a plane and the panel reports callsign, type, registration, height, speed, track, distance, bearing and squawk. Data arrives over Wi-Fi from the free community ADS-B networks adsb.lol and adsb.fi, with automatic failover between them, or from Flightradar24 if you hold a paid API key. Four boards are supported, from the 10 dollar 2.8 inch Cheap Yellow Display to a 1.28 inch round panel and a 3.5 inch Raspberry Pi LCD. Flashing happens in Chrome or Edge over Web Serial, settings go over the same USB cable or the touch screen, and a scriptable USB console exposes everything to automation. The code is MIT licensed with the flasher site, firmware, PC preview and tooling in one repo.
The 3.5 inch build showing the radar scope. Image: OWLOS Flight Radar project page.
Why an ESP32 Can Show Live Air Traffic
Aircraft broadcast their GPS position twice a second on 1090 MHz using ADS-B, unencrypted and receivable with 30 dollars of parts. A worldwide volunteer network of rooftop receivers collects these bursts and republishes them as simple web APIs. OWLOS Flight Radar skips the receiver entirely and asks those APIs over Wi-Fi, which is why a microcontroller with no radio front end can paint a radar picture.
This architecture has three consequences worth understanding. First, coverage follows feeder density, so cities show dense skies and remote areas show little. Second, the board needs internet, unlike a local 1090 MHz receiver that works off grid. Third, anyone can improve the shared picture by feeding, which is where hams with masts come in, covered later. The project picks the pragmatic side of each trade and documents the costs honestly.
The Four Supported Boards
2.8 Inch CYD, the Ten Dollar Classic
The ESP32-2432S028R Cheap Yellow Display pairs an ESP32 with a 320 by 240 resistive touch screen, and one firmware build fits all three screen chips sold under that name. Most boards carry the v2 USB-C screen. A mirrored picture, wrong colors or a negative image means yours is v1 micro USB or v3 ST7789, fixed by pressing BOOT within 8 seconds of power on until the picture looks right. The choice survives resets and power cuts. Readers who already own this board from the CYD Ham Dashboard build can repurpose the same hardware in minutes.
3.5 Inch CYD, the Big Scope
The ESP32-3248S035 upgrades to a 480 by 320 ST7796 panel with resistive or capacitive touch detected automatically. The larger glass lists more planes in the panel, and the firmware puts the second ESP32 core to work drawing half the radar. If you buy one board for this project, this is the sweet spot between price and readability.
1.28 Inch Round CrowPanel, the Wrist Radar
Elecrow CrowPanel combines an ESP32-C3 with a 240 by 240 GC9A01 round display and capacitive touch. The whole glass is the scope. Tap a plane for its card, tap empty glass for hue, look, list and settings, drag the rim to rotate and inside to zoom. It's the most charming option and the least readable, best for desks already full of round gauges.
3.5 Inch RPi LCD Without Touch
A Waveshare RPi LCD (A) on the author OWLOSAirQuality carrier board gives a 480 by 320 ILI9486 display with no touch at all. BOOT steps through planes and held BOOT changes range, with the rest configured from the web page. This target exists because the author already had the carrier hardware, and it proves the display core ports cleanly to new panels.
Flashing and First Setup
Flashing uses ESP Web Tools over Web Serial, so Chrome or Edge on a computer is required. Phones and Firefox can't do it. Connect with a data capable USB cable, press CONNECT and FLASH under your board, pick the port and accept the erase. Missing ports on Windows usually mean the CH340 driver is absent. A board that refuses connection often yields to holding BOOT while plugging in. Touch boards request five calibration crosses once.
Until Wi-Fi lands, the board flies a made up DEMO sky and says so in large letters. The two CYD boards take credentials on their own touch keyboards, complete with ABC capitals, symbols page, delete and show toggles. Every board also accepts settings from the web page over USB, including the round and RPi targets. Privacy handling is thoughtful throughout. Home coordinates round to about a kilometer before any network hears them, the page stores nothing, passwords and API keys enter the board one way and are never displayed or sent back.
Reading the Scope
Three looks cycle under one button. RADAR sweeps a beam over the OpenStreetMap picture, MAP fills the screen north up, CLASSIC renders the same map in OpenStreetMap native colors. Drag the rim to rotate the view until it matches how the screen stands, with snap back to north up. Drag inside to zoom, or step BOOT through 5, 10, 25, 50, 100 and 200 range steps in your chosen units.
Color always means height. Grey sits on the ground, warm tones fly under 10,000 feet, the active theme color covers up to 25,000, white cruises above, and emergency squawks blink for attention. Five themes, Arctic, Phosphor, Amber, Crimson and Synth, recolor the scope without changing the code. Picked planes grow a one minute projection line, with NEXT stepping to the nearest neighbor.
Live traffic with trails and the nearest planes panel. Image: OWLOS Flight Radar project page.
LIST ranks every flight nearest first with type, height, speed, distance and bearing, scrolling under a finger. SET holds range, units, feet or meters, orientation, labels, colors, look, map source, data source, Wi-Fi and touch calibration. The map tiles download once per area and persist in flash. So a previously viewed range redraws instantly with no network at all.
A picked plane with full details and projection line. Image: OWLOS Flight Radar project page.
Where the Planes Come From
The default path costs nothing. The board queries adsb.lol first and falls over to adsb.fi when it goes quiet, both volunteer receiver networks publishing under share alike terms. A paid Flightradar24 API key upgrades answers with origin and destination airports, polled every 20 seconds to conserve credits. Map pictures and town names come from OpenStreetMap under the same open license, with the town lookup using the same rounded home point.
This stack explains the project economics. No receiver hardware, no antenna, no coax, no server bill. The trade is dependence on internet and on volunteers keeping their feeders online, which motivates the ham section below.
For Builders: ESP-IDF Core and a Scriptable Console
The firmware runs ESP-IDF 5.5 with the screen code in core/ written as plain C that also builds a PC preview, so new pages can be drawn and screenshotted on a computer before flashing. BUILD.BAT compiles each of the four screens with its own sdkconfig, GitHub Actions rebuilds everything on every push, and companion tools film tours, bake fonts and annotate guides from the same sources. It's an unusually complete pipeline for a young project.
The USB console deserves special attention. At 115200 baud the board answers OWLOS HELLO with its version, OWLOS GET with a JSON config of network, home, source, theme, uptime and reset history, and OWLOS SET with any subset of fields for scripted provisioning. Passwords and keys never cross back out. That interface turns fleet setup, logging and watchdogs into afternoon Python scripts rather than firmware forks.
The diagnostics page answers the classic embedded question of what happened while nobody watched. It reports uptime, the reset cause among power on, reset pin, watchdog or crash, and the previous run length to the minute, all surviving resets and power cuts.
OWLOS Versus Other Ways to Watch Planes
Item
OWLOS Flight Radar
Raspberry Pi Tracker Display
Phone Flight Tracking App
Hardware
10 to 30 dollar ESP32 board with screen
Pi plus display plus SDR dongle plus antenna
Phone you already own
Receiver needed
No, uses internet aggregators
Yes, own 1090 MHz station
No, uses vendor servers
Works offline
Cached map ranges only
Fully, local reception
Cached areas only
Touch scope UI
Yes, purpose built gestures
Depends on software chosen
Yes, polished commercial UI
Data cost
Free tiers, optional paid key
Free after hardware
Free tier, subscription for extras
Feed the network
No, pure consumer
Yes, shares your reception
No, pure consumer
Openness
MIT, full source and tools
Varies by software stack
Closed
Best for
Desk scope in minutes on pocket money
Hams wanting their own RF station
Travelers checking one flight
Readers who want the full receiving station should see the FlightScnr Pi round display tracker and the mlat.uk multi domain tracking guide. OWLOS and a personal feeder complement each other beautifully, one showing the shared sky anywhere with Wi-Fi, the other hearing it directly off your own antenna.
Ham Radio Tie Ins: Feed, Listen, Script
The strongest ham contribution is feeding. A Raspberry Pi, an RTL-SDR dongle, a 1090 MHz ground plane antenna and free feeder software put your rooftop into adsb.lol and adsb.fi, improving the same picture your OWLOS board shows. Height and clear coax to the antenna beat dongle choice every time, and even a loft antenna hears airliners across a small country.
The scope itself fits shack life well. It draws little power, stays silent, and shows callsigns visitors ask about. The USB console invites automation a shack already runs. Log overhead traffic to the station diary, flash a lamp on emergency squawks, or mirror selected callsigns to the shack dashboard. Receive only operation needs no license anywhere, making it a perfect first build for new licensees and shortwave listeners alike.
Honest Limits
This is a young project with a short commit history. So expect rough edges and answer your own questions from the readable source. Internet dependence means dead zones show only cached maps. Position data lags real time by seconds, fine for spotting, useless for separation. Touch resistive CYDs want firm presses that overshoot small targets. The Flightradar24 key costs real subscription money for origin data most viewers won't miss. And no screen in this price class stays readable in direct sun, so plan shade for window seats.
Sources and Further Reading
OWLOS Flight Radar flasher site and full guide: https://kirindenis.github.io/OWLOS-FLIGHT-RADAR/
Firmware source, boards, tools and README: https://github.com/KirinDenis/OWLOS-FLIGHT-RADAR
Free community ADS-B network, primary source: https://adsb.lol/
Free community ADS-B network, failover source: https://adsb.fi/
Flightradar24 API for key holders: https://fr24api.flightradar24.com/
Browser flashing via ESP Web Tools: https://esphome.github.io/esp-web-tools/
Espressif ESP-IDF framework docs: https://docs.espressif.com/projects/esp-idf/
OpenStreetMap copyright and license: https://www.openstreetmap.org/copyright
CYD Ham Dashboard repurposing the same 2.8 inch board: https://hamradio.my/esp32-cyd-ham-dashboard-hamclock-touchscreen/
Raspberry Pi ADS-B tracker display build: https://hamradio.my/flightscnr-pi-round-display-adsb-flight-marine-tracker-raspberry-pi/
Live air, maritime and space tracking guide: https://hamradio.my/mlat-uk-live-air-maritime-space-tracker/
73 de 9M2PJU
The post OWLOS Flight Radar: Live ADS-B Air Traffic on a $15 ESP32 Display appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/owlos-flight-radar-esp32-adsb-scope/

#owlosflight #flightradar #airtraffic #display #yellowdisplay #inch #boards #usb #cheapyellow #cyd

#rss #feeds #linksharing #qwibbit

Traccar for Hams: Private Live GPS Tracking Without Touching APRS ...

2026-11-01T17:08:15Z

Traccar for Hams: Private Live GPS Tracking Without Touching APRS

What Traccar Is
Traccar is a free and open source GPS tracking platform that collects position reports from phones and dedicated LTE trackers over the internet and shows them on a live self-hosted map with trips, geofences, alerts and reports.
If you have ever watched a club convoy disappear into town traffic, waited for a support driver who took a wrong turn on the way to a SOTA trailhead, or tried to explain to family where the EmComm team is during a public service event, you already know the problem Traccar solves. You want dots on a map, names next to the dots, a record of where everyone went, and a ping when someone arrives. You don’t always want to put that traffic on 144.390 MHz.
APRS is brilliant at what it was built for. It’s also public by design, limited by RF coverage and digipeater density, throttled to 1200 baud packet bursts, and tied to your callsign and license. Traccar takes a different path. It moves the tracking to IP. A phone app or a small LTE tracker sends GPS over cellular data or Wi-Fi to your own server. Only the people you invite can see it. Update rates are measured in seconds, not minutes. There’s no path setting, no digipeater alias, no hidden transmitter hunt when a tracker goes quiet because it drove behind a hill with no iGate nearby.
This guide covers what Traccar actually is under the hood, where it beats APRS for ham use, where APRS still wins, and how to set up a practical station: a Docker server on a cheap VPS, phones running Traccar Client, an optional hardware tracker in a club vehicle, geofences around base camp and the summit car park, and alerts that go to email and push instead of an APRS message that may never arrive.
Traccar web dashboard with live device positions, status list and map layers. Image credit: Traccar project, Apache 2.0.
A Short History of a Quiet Workhorse
Traccar started in 2009 as the solo project of Anton Tananaev and grew the slow way, one GPS protocol at a time. That matters because GPS tracking hardware is a fragmented mess. Every fleet vendor ships a slightly different TCP or UDP dialect, a different login handshake, a different heartbeat. A tracker platform lives or dies on how many of those dialects it speaks.
Seventeen years later the numbers tell the story. The core repo at https://github.com/traccar/traccar holds more than 10,000 commits, around 8,000 stars and 3,500 forks. The Java server speaks more than 200 protocols covering more than 2,000 device models, from five dollar GT06 clones to professional Teltonika and Quectel units with ignition sense, fuel probes and 1-Wire temperature sensors. The license is Apache 2.0, which means a club, a scout group or a small business can run it without a license server calling home.
The project split into clear parts early and kept that shape. The traccar repo is the Java backend that ingests device connections on ports 5000 to 5300, stores positions in H2 for testing or MySQL, PostgreSQL and TimescaleDB for real use, and serves a REST API. The web dashboard lives in traccar-web. Phone viewing is handled by Traccar Manager. Phone tracking is handled by Traccar Client for Android and iOS. There is even a Client SDK for building the same background engine into your own Flutter, React Native or native app.
For hams, that history means two practical things. First, you aren’t betting an event on a weekend project that vanishes next year. Forums at https://www.traccar.org/forums/ go back more than a decade and the demo server at https://www.traccar.org/demo-server/ is always up for testing. Second, if your club already owns a box of old fleet trackers, there’s a good chance Traccar already speaks their language.
Why Hams Keep Reaching for Something Besides APRS
Nobody here needs APRS explained, but it helps to name exactly where it pinches when you just need to keep track of people.
APRS transmits in the clear on shared VHF. In the United States that means 144.390 MHz, in Malaysia 144.525 MHz, in much of Europe 144.800 MHz. Every position is public on APRS-IS and sites like aprs.fi within seconds. That openness is the point for tactical callsign visibility, and it is exactly what you don’t want when you’re tracking scouts, family members, or the exact location of a served agency shelter during an exercise.
Coverage ends where the RF network ends. A valley with no digipeater, a town with a deaf iGate, a 5 watt tracker buried in a car footwell, and your dots stop. The classic fixes all cost RF effort: more digipeaters, better antennas, fill-in nodes, careful path settings like WIDE1-1 versus WIDE2-1. On a busy channel you also share 1200 baud with everyone else, so responsible trackers beacon every few minutes, not every few seconds.
Then there is the hardware tax. A proper APRS tracker needs a radio, a TNC or a modem chip, a GPS, and power wiring that survives engine cranking. A Mobilinkd plus an HT works, a TinyTrak plus a mobile works, a purpose built APRS HT like a Kenwood TH-D74 works, but none of it is as cheap or as simple as an app you already carry. Receiving is easy. Transmitting legally still needs a license, which rules out tracking non-ham volunteers, youth group helpers, or the driver your club hired for the day.
Traccar flips each of those constraints. Coverage follows cellular coverage, which along highways and in towns is usually better than digipeater coverage. Privacy follows your server, not the whole internet. Update intervals follow your data plan, with 5 to 15 second live tracking that looks smooth instead of jumpy. Participants need a phone, not a license. The cost is dependence on commercial networks and a server you keep running, which is why the honest answer isn’t Traccar versus APRS forever, but the right tool for the outing.
Use Traccar when the team has cell signal, privacy matters, you want second level updates and history, or half the people you need to track aren’t licensed. Stay with APRS when you are truly off grid, when you want RF resilience after infrastructure fails, or when public visibility on APRS-IS is itself the goal, like a high altitude balloon flight the whole community follows.
Traccar Versus APRS at a Glance
Item
Traccar
APRS
Transport
Cellular data or Wi-Fi to your server over TCP, UDP or HTTP
VHF packet radio at 1200 baud to digipeaters and iGates
License needed to be tracked
No, any phone or LTE tracker works
Yes for RF transmission, callsign required
Coverage
Wherever phones have data, worldwide roaming with the right SIM
Wherever digipeaters and iGates hear you
Privacy
Private by default, you invite viewers per device and group
Public by design, gated to APRS-IS within seconds
Update rate
5 to 30 seconds typical, smooth live trails
1 to 10 minutes typical to protect the shared channel
Maps and history
Modern web map, satellite layer, trips, stops, charts, Excel export
aprs.fi and clients, great for RF path insight, thinner reporting
Alerts
Geofence enter and exit, speeding, SOS, offline, ignition, fuel, maintenance, push plus email plus SMS
Messaging and bulletins, no built-in geofence alerting
Hardware
Phone app is enough, optional 20 to 80 dollar LTE trackers
Radio plus modem plus GPS, or APRS capable HT or mobile
Failure mode
Dies with cell outage or dead server
Dies with RF shadow or dead digipeater, survives cell outage
Cost shape
VPS at 5 to 10 dollars a month plus data SIMs, software free
No server bill, hardware cost per tracker higher
A third row many hams ask about is Meshtastic over LoRa. It is closer to APRS in spirit, off grid, license free on ISM, encrypted if you want, but hop limited and low throughput. Think of it as tactical text and coarse position for the field team, while Traccar is the detailed live picture at net control where internet exists.
Under the Hood: Server, Apps, Protocols and Ports
It helps to see Traccar as four layers that you can adopt one at a time.
The server is Java, multithreaded, built with Netty for the hundreds of concurrent device sockets. It listens on a wide default range, TCP and UDP 5000 to 5300, with each protocol pinned to its own port. A GT06 tracker might report to 5023 while a Teltonika unit reports to 5027. The web UI and API sit on 8082 by default behind whatever reverse proxy you choose. Positions land in the database, events are evaluated in memory, and notifications fan out through configured channels. The full architecture is documented at https://www.traccar.org/architecture/ and the API at https://www.traccar.org/traccar-api/.
The database choice is the one decision to get right on day one. The built-in H2 file is fine for an afternoon test with three phones. For a club server that runs all year, use MySQL or MariaDB for a small install, or PostgreSQL with TimescaleDB for a large one with months of history. Timescale turns huge position tables from sluggish to instant because time series queries are its home turf. The project docs walk through both at https://www.traccar.org/mysql/ and https://www.traccar.org/postgresql/.
The web app is what net control stares at. Live map with road and satellite layers, device list with online status, per device sensor panels, geofence editor, trips and stops timeline, route replay slider, reports with charts, calendars for quiet hours, drivers and groups for permissions, and computed attributes that turn raw tracker fields into useful numbers like gallons or volts. Reports export to Excel, which served agency after action reports love.
Traccar Manager for Android and iOS is the viewer for people in the field. Traccar Client for Android and iOS is the tracker itself. Install Client on a volunteer phone, type in your server address and a numeric device identifier, set the desired interval and distance filter, and the phone starts reporting in the background with battery aware location APIs. The default server in a fresh install points at the public demo, which is perfect for a five minute test before you point it at your own box. Setup details are at https://www.traccar.org/client/.
Protocols are where the 200 number matters. If you buy hardware later, check the supported device list at https://www.traccar.org/devices/ before you buy. Teltonika FMB series, Quectel GV series, Concox GT06 and JT701, Xexun TK series, Meitrack, CalAmp and many more each have a named protocol handler and a default port. The server identifies most trackers by IMEI or a serial ID printed on the label, so provisioning is usually typing that number into the web UI once.
Ham Use Cases That Fit Naturally
Field Day and contest support runs smoother with live dots. Put Client on the shuttle drivers between parking, site and food runs. Net control sees ETAs without voice check ins every ten minutes. The voice net stays free for real traffic.
SOTA and POTA activations gain a safety layer. The activator carries a radio, the support driver at the trailhead carries a phone running Client. If the activator is overdue, you have a last known road position, a track into the site, and timestamps, all private to the club group. No need to broadcast the parking spot of an expensive vehicle on a public map.
Public service events are the strongest fit. Marathons, charity rides, parades and search exercises involve a mix of hams and non-hams, moving checkpoints, sweep vehicles and supply vans. Traccar gives the event organizer one private screen with every vehicle, geofences around each checkpoint for automatic arrival logs, and speeding alerts when a tired driver pushes too hard on closed course roads. APRS can still cover the ham core on RF while Traccar covers everyone else on IP, and net control watches both.
EmComm groups can use it for readiness without drama. Weekly vehicle checks become a trip report. Generator trailers get a small wired tracker with ignition and voltage sensing, so you know if shore power drops. The key discipline is honesty in the comms plan: Traccar is an internet system and it isn’t the backup when the internet fails. Write it as the day to day coordination layer, keep RF methods as the degraded mode layer.
Family and club youth outings benefit most from privacy. A scout hike, a lighthouse activation with kids, a convoy to a hamfest in another state, all work with invited viewers only. Share links expire, accounts are per event, and nothing lands on a public aggregator unless you deliberately publish it.
Vehicle and trailer security is a quiet bonus. A 30 dollar magnetic LTE tracker tucked in a tower trailer or a club go box reports once a minute and screams if it moves at 2 AM. Pair it with a geofence around the storage lot and an SMS alert to two trustees. Recovery beats paperwork.
What to Buy and What to Skip
Start with phones. Every modern Android and iPhone already has a better GPS, a bigger battery and a more reliable data stack than a bargain tracker. For a first event, Traccar Client on five phones teaches you more than any hardware purchase.
When you outgrow phones, buy one proven wired tracker before you buy ten wireless ones. The Teltonika FMB920 and FMB125 show up in fleet use for a reason: wide voltage range, clean ignition detection, internal battery backup, solid antennas, and first class Traccar support. For magnetic portable use, a Concox or Quectel LTE unit with a large internal cell and USB-C charging is easier to live with than the old TK103 clones that needed a special cable and a prayer.
Skip anything that only speaks a vendor cloud app with no raw server setting. If the configuration tool has no field for a custom server IP and port, Traccar can’t use it. Skip 2G only trackers as well. Carriers in many countries have already shut 2G down, and a tracker that can’t register is a paperweight. Buy LTE Cat-M1, NB-IoT or LTE Cat-1 depending on local carrier support, and test one SIM from the actual carrier you’ll use at the venue.
SIM choice matters more than tracker brand. A full data plan is overkill for a tracker that sends a few kilobytes per report. IoT data SIMs, low cost prepaid data SIMs, or a shared pool plan all work. Test indoors, in a car footwell, and at the actual site. Metal roofs and heated windscreens kill GPS far faster than any software setting.
Setting Up a Club Server in an Afternoon
The fastest durable setup is Docker Compose on a small VPS with MySQL, behind HTTPS. The official VPS guide at https://www.traccar.org/install-vps/ and Docker notes at https://www.traccar.org/docker/ are the reference. What follows is the ham shaped version that works for a club with one Linux volunteer.
Pick a 1 CPU, 1 to 2 GB RAM VPS in the region closest to your users. Install Docker Engine with Compose using the official script from Docker, disable the userland proxy to avoid a process per tracker port, create persistent directories for data and logs, drop in the official MySQL compose file from the Traccar repo, and bring it up. The web UI appears on port 8082. The full port range 5000 to 5300 needs to reach the host for hardware trackers, but phones reporting over HTTP only need the web port, so you can open the tracker range later when hardware arrives.
Put HTTPS in front before you invite anyone. The secure connection guide at https://www.traccar.org/secure-connection/ shows the Caddy pattern that fetches a free certificate automatically. This isn’t optional polish. Phones send credentials and positions over this link, and browsers now punish plain HTTP with warnings and broken geolocation.
Log in as admin, change the admin password immediately, create a group per event such as Spring Bike Ride 2026, and create users for net control, checkpoint leads and viewers with only the permissions they need. Create devices with the identifier shown in Traccar Client on each phone, or the IMEI for hardware trackers, and place each device in the right group. Permissions and groups are explained at https://www.traccar.org/permissions-groups/.
Draw geofences next. A circle around base camp, polygons around each checkpoint, a corridor along the route if the course is complex. Link each geofence to the relevant group, not to individual devices one by one, so next year you reuse the same shape with new callsigns. The geofence docs at https://www.traccar.org/geofences/ cover circles, polygons and polylines.
Build two notifications before the event, not during it. One for geofence enter and exit

https://hamradio.my/traccar-gps-tracking-amateur-radio-without-aprs/

#gpstracking #traccar #aprs #workhorsetraccar #lte

#rss #feeds #linksharing #qwibbit

OpenOBD: A Free Offline OBD-II Scanner for Hams What OpenOBD Is ...

2026-11-01T16:32:41Z

OpenOBD: A Free Offline OBD-II Scanner for Hams

What OpenOBD Is
OpenOBD is a free and open source Android app that turns a cheap ELM327 Bluetooth adapter into a full OBD-II workshop companion with live gauges, trouble code reading and clearing, emissions readiness, vehicle identity data, a service schedule and a raw command terminal, all with no network permission, no account, no ads and no analytics.
The most interesting car tool a radio amateur owns is the car itself. Summits don’t come to you. Field Day sites, SOTA trailheads, POTA parks, EmComm deployments, club foxhunts, hamfest road trips, all of them start with a vehicle that has to get there and back. A check engine light the night before a 5 AM convoy is a special kind of dread, and the dealer wants your morning plus a diagnostic fee to tell you what a 25 dollar adapter and a free app can tell you in the driveway. OpenOBD exists for exactly that moment, and it respects the same values hams already hold: open protocols, hardware you own, software that works offline, and zero telemetry phoning home.
The project lives at https://github.com/Foadsf/OpenOBD, GPL-3.0 licensed, written in Kotlin with Jetpack Compose and Material 3, and built around one named reference adapter, the Veepeak OBDCheck BLE and BLE+. It’s a young project, three commits at the time of writing, and refreshingly honest about what generic OBD-II can and can’t do. Its docs folder even ships a verification review that grades the author’s own research against the vendor manual and the SAE standards, claim by claim. That kind of candor is rare in car software, and it’s why this app deserves a close look even in its early days.
OpenOBD’s five screens: Live gauges, Codes, Vehicle, Care schedule and raw Terminal. Screenshots: Foadsf OpenOBD, GPL-3.0.
Why Your Vehicle Is Part of Your Station
Ask a seasoned activator what failed on their worst outing and the answer is rarely the radio. It’s the support vehicle. A flat battery at the trailhead, an overheating climb up a forest road with a loaded trailer, a mystery limp mode halfway to Field Day with a tower trailer behind you. The radio shack gets checklists, SWR checks and backup rigs. The vehicle that carries it all gets fuel and hope.
An OBD-II scanner closes that gap for pocket money. Before a long drive to a rare grid or an EmComm exercise, five minutes with a scanner answers the questions that matter. Are there stored or pending fault codes even though the dash looks clean. Is the charging system healthy. Are the emissions monitors ready or did someone recently clear codes to hide a problem on a used 4×4 you are about to buy for portable work. What is the actual coolant temperature on a grade, not the damped needle on the cluster that sits centered from 75 to 110 degrees.
For club vehicles and EmComm trailers the case is even stronger. A club van or a towed communications trailer sits for weeks, then gets asked to perform on exercise day. A standing practice of plugging in monthly, logging battery voltage and codes, and keeping the service schedule honest turns breakdown roulette into boring reliability. OpenOBD keeps that service schedule on the phone, reads the odometer from the car where the car reports one, and exports a plain text report you can hand to the club trustee or the mechanic.
There is a cultural fit too. Hams are the people who read the manual, learn the protocol, and distrust black boxes that require a cloud account to show you your own data. OpenOBD is built like ham software. It speaks a documented protocol to hardware you can buy anywhere, it works with the internet switched off, and when you want to go deeper there’s a terminal waiting for you.
A Short History of the Port Under Your Dash
On board diagnostics didn’t start friendly. In the 1980s every carmaker had its own blink codes, proprietary connectors and secret readers, and independent workshops were locked out. California forced the issue. The California Air Resources Board required standardized diagnostics, OBD-I, on 1988 and later cars sold in the state, because emissions controls are useless if nobody can verify them in the field.
OBD-II was the real settlement. Required on all 1996 and newer cars and light trucks sold in the United States, later adopted as EOBD in Europe and JOBD in Japan with regional timelines, it standardized the 16 pin diagnostic connector (SAE J1962), the service and parameter definitions (SAE J1979), the fault code format (SAE J2012, the familiar P0xxx powertrain codes), and the physical signaling layers. Suddenly one tool could talk to any compliant car, at least for engine and emissions data. That compromise, standardize the emissions core and leave the rest proprietary, still defines what generic scanners can see today.
The ELM327 made that standard accessible to mortals. Elm Electronics built a microcontroller that translates between the car’s alphabet soup of protocols, J1850 PWM and VPW, ISO 9141-2, ISO 14230-4 KWP, ISO 15765-4 CAN at 11 and 29 bit, and a simple ASCII command set over serial that any programmer can drive. Type 010C and get engine RPM back. The chip’s AT style command set became the de facto language of cheap car diagnostics, and a flood of Bluetooth adapters followed. Then came the clone epidemic. Countless bargain adapters claim to be ELM327 v1.5 or v2.1 while running incomplete firmware that chokes on multi frame responses or lies about its version. Anyone who has fought a no name adapter knows the pain, and it is exactly why OpenOBD includes an adapter check screen that probes which ELM327 commands the hardware really answers, version by version from v1.0 to v2.2.
Bluetooth Low Energy changed the daily experience. Classic Bluetooth adapters need system pairing with a PIN, usually 1234, and they fight with car stereos for the connection. BLE adapters like the Veepeak OBDCheck BLE need no pairing at all on Android, connect straight from the app, sip power, and leave your phone’s classic slot free for the hands free kit. That’s the adapter OpenOBD is built and tested around, with classic adapters still supported for people who already own one.
What OpenOBD Actually Does
Five screens plus a simulator cover the whole generic OBD-II job. Nothing here requires a subscription, and everything works in airplane mode with Bluetooth on.
Live shows gauges you pick plus every sensor the car reports. Under the hood this is service $01, real time powertrain data, with around 100 standard parameter IDs decoded, RPM, speed, coolant temperature, manifold pressure, fuel trims, intake air temperature, timing advance, oxygen sensor voltages, battery voltage and more. Anything the car reports that has no standard decoder is shown raw rather than hidden, which is the right call for oddball vehicles. Whatever is on screen can be recorded to CSV, so you can log a climb to a summit site and review coolant temperature and fuel trims later at home.
Codes handles the check engine light properly. Stored codes ($03), pending codes ($07) that have failed once but not yet lit the lamp, and permanent codes ($0A) that can’t be erased with a scan tool and only clear when the car verifies the repair itself. Each code gets a plain language description from a built in database of around 230 common generic codes, plus the system it belongs to and a one tap web lookup for the rest. The freeze frame ($02), a snapshot of engine conditions when the fault set, comes along for context. Emissions readiness shows both since codes were cleared and for the current drive cycle, so you know whether the car has completed its self tests. Clearing ($04) sits behind a guarded confirmation, as it should, because clearing also wipes readiness and can make a car fail inspection until it retests. The app says so in the dialog instead of letting you learn it at the inspection lane. A finished diagnosis can be shared as a plain text report for your mechanic, no screenshots required.
Vehicle collects identity and calibration data. The VIN, every control unit’s name, calibration IDs and calibration verification numbers ($09), the car’s own on board monitor test results with pass or fail limits ($06), odometer and distance and time since codes were cleared where the car reports them. The adapter card deserves special mention. It shows connection details, the BLE layout discovered at runtime, and the adapter check that exposes clone firmware pretending to be something it isn’t. Protocol choice and metric or imperial units round out the screen.
Care is a service schedule by distance and time. Oil, filters, plugs, brake fluid, coolant, plus your own custom items. Defaults are editable starting points, with the owner’s manual named as the real authority, which is the honest way to ship maintenance intervals. Where the car reports an odometer over PID $A6 the app can use it, otherwise you type it in. For a club van with rotating drivers, one phone holding the schedule beats a windshield sticker nobody reads.
Terminal is the hacker’s door. Send any AT or OBD command straight to the adapter, for example selecting a header with ATSH7E0 then asking for manufacturer data with 22F190, and the app restores its own adapter settings afterwards so you can’t strand the session in a weird state. This is how an ELM327 exposes the good stuff to people who know their car’s extended PIDs, and its presence signals what kind of project this is. Nothing is dumbed down, the sharp tools are just kept in a clearly labeled drawer.
Demo runs a simulated car with no adapter at all. Every screen works against the simulator, which doubles as the project’s test rig. New users can learn the whole app on the couch, and every OBD service is exercised by unit and end to end tests against that same simulated adapter. If you are curious but adapter poor, try the demo first.
Privacy as a Feature, Not a Slogan
Here is the line from the project page that matters most: no network permission, no account, no analytics, no ads, and continuous integration fails any build that adds the INTERNET permission. On Android that is checkable, not marketing. Open the manifest and the permission is absent, so the app physically can’t exfiltrate your VIN, your location history or your fault codes no matter how much it might want to.
Compare that with the mainstream. The popular OBD apps are free because you are the product pipeline, accounts before scanning, location and analytics SDKs bundled in, driving data synced to vendor clouds, feature gates and subscription tiers for the codes that matter. Some of that funds genuinely good manufacturer specific coverage, which is a fair trade many people choose knowingly. But for reading your own car’s generic data, none of it is necessary, and OpenOBD proves it by doing the whole job in a 1.6 MB package with zero dependencies beyond the UI toolkit.
Hams already live this argument. We run offline loggers for Field Day, we keep paper duplicates of critical records, we choose tools that work when the cell network doesn’t. A diagnostic app that works in a concrete parking garage, on a mountaintop with no bars, or across a border with no roaming isn’t a philosophical statement. It’s the tool that works where vehicles actually break.
OpenOBD Versus the Usual Suspects
Item
OpenOBD
Torque Pro
Car Scanner
Branded dongle apps (BlueDriver, FIXD)
License and source
GPL-3.0, fully open
Proprietary
Proprietary, free tier plus paid
Proprietary, tied to vendor dongle
Price
Free
Paid app
Free with limits, paid pro
Free app, requires 50 to 120 dollar dongle
Account, ads, analytics
None, no network permission at all
No account, minimal permissions historically
Account prompts, ads in free tier
Account required, cloud synced history
Works fully offline
Yes, everything on the phone
Yes for core gauges
Mostly, some features call home
No, cloud dependent
Generic OBD-II depth
Strong: $01 to $04 plus $06, $07, $09, $0A, freeze frame, readiness
Strong plus custom PIDs and plugins
Very strong plus graphing
Strong but locked to their hardware
Manufacturer modules (ABS, airbag, coding)
No, generic engine and emissions only
Some via plugins and custom PIDs
Yes, paid packs for many brands
Yes, the reason to buy them
Adapter flexibility
Any ELM327, tuned for Veepeak BLE
Any ELM327 or compatible
Any ELM327 or compatible
Their dongle only
App size and weight
About 1.6 MB
Tens of MB
Tens of MB
Tens of MB
The honest read: Torque and Car Scanner do more, especially on dashboards, graphing and manufacturer specific data, and a branded dongle earns its price when you need airbag or ABS access on a specific car. OpenOBD wins on trust, weight and price for the generic job. It’s the app you hand to a new driver, install on a club tablet, or keep on a head unit without wondering what it uploads. Use the right tool per job, and let OpenOBD own the everyday checks.
Know the boundary before you buy parts. Generic OBD-II covers engine and emissions. ABS, airbags, body modules, TPMS details, transmission temperature on most cars, DPF specifics, service light resets and bi directional active tests live behind manufacturer protocols and need brand specific tools. OpenOBD states this plainly instead of implying otherwise, and its review doc lists whole protocol families the reference hardware doesn’t speak: heavy duty J1939, Ford and Mazda MS-CAN, GM SW-CAN, OBD-I, motorcycles, and possibly very new WWH-OBD cars. That candor is a feature. A tool that tells you where its map ends is one you can trust where the map is drawn.
Under the Hood: Kotlin, Compose and a Simulated Car
The codebase is small enough to read in an evening, which for open source is a compliment. Three packages do the work. The elm package holds the Bluetooth LE and Classic transports, the ELM327 command link, frame parsing and the demo simulator. The obd package implements the SAE J1979 services, the PID decoders, the fault code tables, readiness logic and monitor test parsing. The ui package is Jetpack Compose screens in Material 3. Kotlin throughout, no other dependencies, every library version pinned in the Gradle catalog.
Two engineering choices show experience. First, the ELM327 link keeps exactly one command in flight and waits for the > prompt instead of sprinkling fixed delays between queries, because the adapter is half duplex and overflow comes from monitor modes, not from impatience. Second, initialization leaves headers on and reassembles ISO-TP multi frame responses per ECU itself, instead of the common shortcut of headers off that makes multi ECU and multi frame answers indistinguishable. BLE writes are chunked to 20 bytes to dodge MTU bugs on cheap adapters, and the GATT layout is discovered at runtime across the known UUID families rather than assumed, which is why the app can show you what your specific adapter actually exposes.
Testing is stronger than the commit count suggests. Sixteen unit and end to end tests run the whole stack, ISO-TP framing, the ELM327 link and every OBD service, against the simulated adapter in DemoTransport.kt. The build needs Java 17 plus an Android SDK with the 37.0 platform and build tools, compiles down to targetSdk 35 with minSdk 23 covering Android 6 and newer, and produces a minified release APK around 1.6 MB. Plain SharedPreferences and JSON hold settings instead of a database, a deliberate choice that keeps the app tiny. GitHub releases are signed with the maintainer certificate, F-Droid builds carry its own signature, and self built APKs use your debug key, so Android will ask you to uninstall before switching between signatures. Standard Android behavior, worth knowing before you mix sources.
Using It in the Field
The field routine is short enough to laminate. Plug the adapter in until its blue light comes on. Switch the ignition on, engine off is fine for code work. Open OpenOBD, grant the Nearby devices permission, and tap the adapter under the Bluetooth LE list. No pairing step, that’s the point of BLE. If yours is a Classic only adapter, pair it first in Android Bluetooth settings with PIN 1234, then pick it under the paired list. Remember one app owns the adapter at a time, so close Torque or Car Scanner first or you’ll stare at a connection error that isn’t a bug.
Start with Vehicle and run the adapter check on any new or borrowed dongle. Clones that cla

https://hamradio.my/openobd-free-offline-obd-ii-scanner-elm327-android/

#free #openobd #hams #vehicle #scanner #fieldday #codes

#rss #feeds #linksharing #qwibbit

MeshLive: Encrypted Voice Calls, Video Calls and Livestreams Over ...

2026-10-30T19:06:30Z

MeshLive: Encrypted Voice Calls, Video Calls and Livestreams Over Bluetooth Mesh with No Internet

TL;DR: Quick Summary
MeshLive is a free and open source GPL-3.0-or-later Android app that carries encrypted voice calls, video calls, text chats and livestreams over a Bluetooth mesh with no internet, no SIM card, no account and no server. Every phone running it can relay: a livestream hops from phone to phone so range grows with the crowd, and calls go direct or through up to four relays. It uses three Bluetooth transports (LE GATT pairing-free by default, Classic for throughput, LE L2CAP CoC experimental), AES-256-GCM link encryption plus a second end-to-end layer, rotating Bluetooth IDs and QR-verified contacts. Install comes from GitHub releases or Obtainium with certificate-pinned verification, and the APK can even be shared phone to phone with no internet at all. The app is in beta with about 300 stars and 44 forks: video is small, audio-only mode reaches furthest, and the authors warn against relying on it where safety depends on it. For amateur radio operators it’s a natural companion for EmComm, Field Day, SOTA and POTA groups, foxhunts, club events and internet-shutdown scenarios: short-range team voice plus live video from the antenna site, all off grid.
What Is MeshLive?
MeshLive is a free open source Android application that provides end-to-end encrypted voice calls, video calls, messages and livestreams between nearby phones over a Bluetooth mesh network with no internet connection, SIM card, account or server involved.
The project lives at https://github.com/MeshLive/meshlive-android with sibling builds for Windows (beta) and Apple platforms (source build, experimental). Bluetooth alone reaches about 10 to 30 metres; MeshLive stretches that by turning every phone running it into a relay, so a stream or call hops from phone to phone and the usable footprint grows with each participant.
How MeshLive relaying works: four phones spaced about 20 metres apart pass a livestream along, stretching reach to about 60 metres and growing with every phone. Image credit: MeshLive project, GPL-3.0-or-later.
Project Snapshot
Item
Detail
Project
MeshLive for Android by the MeshLive org
License
GPL-3.0-or-later, free and open source
Requirements
Android 10 or newer, Bluetooth, same app on both ends
Status
Beta: works on tested phones, do not rely on it where safety depends on it
Traction
About 300 stars, 44 forks, 32 commits at write time
Transports
LE GATT default (no pairing), Bluetooth Classic (paired, higher throughput), LE L2CAP CoC (experimental); discovery always over Bluetooth LE
Media
Live video in AV1, VP9, HEVC or H.264 with mono 48 kHz Opus audio at 32 kbps; audio-only radio mode available
Encryption
Per-phone signing keys, signed handshakes, ordered AES-256-GCM records plus a second end-to-end layer relays cannot read
Distribution
GitHub releases per-ABI APKs, Obtainium auto-updates, phone-to-phone offline APK share, Windows beta, Apple source builds
Links
Android repo, latest release, user guide, privacy doc, supported phones
What It Does: Calls, Livestreams, Contacts, Privacy
Calls
Voice and video calls with people nearby, end-to-end encrypted phone to phone. Missed-call history with call back, block-unknown-callers switch, hold-Mute-to-talk intercom style voice, ringer plus Do Not Disturb handling, proximity screen-off lock, answer and decline from the notification and over the lock screen. A GATT call that loses one audio direction can re-link itself on a fresh Bluetooth link within about 12 seconds without redialing, keeping the same keys. Relayed calling through up to four relays exists as an experimental debug feature.
Livestreams
Go live and share the phone camera with everyone around, or go audio-only like a local radio station that keeps playing with the screen off. Viewers can relay the stream onward (on by default, auto-paused below 20 percent battery or in Battery Saver), and each receiver verifies the stream truly comes from the original source. Authenticated pause, resume and stop states keep control with the broadcaster. Because Bluetooth is slow, video is small; audio-only streams use a fraction of the capacity and reach the most phones.
Contacts
Add people by scanning each other’s QR code in person or comparing a safety code, then call them from Contacts whenever they are nearby. Strangers never see your real name: they see a random two-word label such as Gold Fox, or a tilde name they asserted themselves, while your chosen name travels encrypted only to people you call. One-time five-minute codes and mutual scanning keep verification honest.
Privacy by default
No internet path for media, no SIM requirement, no account, no ads, no analytics, no servers. Rotating Bluetooth IDs, identity-hiding call handshakes and per-stranger throwaway identities are on by default, plus a Hide phone name default that renames the handset to Phone so nearby scanners learn nothing. Everything can be wiped from inside the app with a typed DELETE confirmation. Experimental store-and-forward messages (text, voice notes up to two minutes, photos, files to 64 MB) wait for accepted contacts and deliver on next proximity.
How It Works Under the Hood
Each hop adds a little delay and Bluetooth airtime is spent twice per forward, so quality falls with chain length: a multi-phone relay test in October 2026 showed relayed viewers at 0.9 to 7.6 FPS at 300 kbps. Short chains of low-rate video are realistic; long chains of high-rate video aren’t. Auto video quality starts near 360p on a typical link and adapts up or down, dropping to audio-only on very weak links rather than freezing.
Transports trade reach against speed: LE GATT works pairing-free on the most phones but is slowest, Classic needs a one-time mutual pairing in Android Bluetooth settings and carries the most, LE CoC is experimental with stack bugs on some phones. Discovery is always Bluetooth LE. A device-family policy registry centralizes known Bluetooth-stack and codec quirks with per-entry kill switches.
Install, Update and Verify
Install
On the phone open the latest release and tap the APK (pick MeshLive.apk arm64 for almost every phone since 2016, armeabi-v7a for older 32-bit models, universal when unsure).
Allow the browser to install unknown apps, then tap Install.
Open MeshLive and grant Nearby devices, Camera and Microphone, then enable Bluetooth.
Updates three ways
Obtainium: add the repo URL in Obtainium for automatic release tracking.
In-app check: opt in once to Check for updates when online; the app then queries the GitHub release API at most daily, downloads nothing until tapped, and installs only same-key signed, newer, matching-ABI builds whose SHA-256 matches the published sums. Android always confirms; nothing installs silently or downgrades.
From a nearby phone: connected peers advertise their version, and Settings can share the installed APK over Quick Share, Nearby Share, Bluetooth or experimental Wi-Fi Direct, with on-screen APK and certificate hashes to compare.
Verify a release
The release signing certificate SHA-256 is published in the release notes and the repo:
2E:EF:C0:5A:44:45:22:52:9D:9F:75:D7:EB:CE:FD:08:86:47:07:F2:1A:79:04:1D:B5:1E:BF:F8:36:49:E2:13
Check with apksigner verify --print-certs (single matching signer), sha256sum -c SHA256SUMS, and gh attestation verify for the source manifest. Never mix APKs from different publishers on one phone; updates require the same signing key.
Daily use notes
Tabs are Live, Calls (Nearby plus Recent), Contacts and Settings. First start asks to hide the phone Bluetooth name (recommended default) and offers one-tap background-running allowance so relays keep working with the screen off (relaying costs noticeable battery and warmth). Android 10 and 11 need the Location switch on for Bluetooth discovery even though the app never reads location. Contacts verify in person; everyone else stays an ID or rotating nickname.
How Amateur Radio Operators Can Use MeshLive
1. EmComm when the grid and internet fail
Floods, storms and outages routinely take out cellular data first while handsets stay powered. A team with MeshLive preinstalled keeps short-range voice coordination, audio-only broadcast announcements from net control, and contact-verified calling with no tower, no Wi-Fi and no account server. Pre-share the APK and verify contacts before deployment day, because there may be no internet to fetch it later.
2. Field Day and portable site talk-around
Field Day sites sprawl across tents, masts and parking. MeshLive gives tent-to-tent voice without tying up an amateur frequency for logistics chatter, plus live video from the antenna crew back to the operating tent (for example showing a masthead during raising). Audio-only mode is the workhorse here: lowest bandwidth, longest reach, plays with the screen off in a pocket.
3. SOTA, POTA and summit groups
Activators split across a summit (operating position, antenna ledge, parking shuttle) stay in voice contact over Bluetooth relay even where cellular coverage drops behind the ridge. The livestream lets the logger watch the antenna tune-up in real time. Remember the 10 to 30 metre per-hop Bluetooth reality: this suits one summit group, not valley-to-peak links. RF still does the QSOs; MeshLive does the logistics.
4. Foxhunts and direction-finding events
Hunt organizers can broadcast audio-only instructions and boundary changes to all teams at the start line, then take voice check-ins without burning a repeater channel. Because strangers see only rotating IDs, teams keep callsign privacy until they choose to identify, matching good hunt practice.
5. Club nights, hamfests and exam sessions
In a noisy hall with saturated Wi-Fi, MeshLive streams a demo bench to nearby phones and takes questions by voice call. VE teams can coordinate candidate flow across rooms with no venue network dependency. The rotating stranger IDs plus block-unknown-callers switch keep unsolicited calls out during presentations.
6. Internet-shutdown and censorship resilience
Where authorities throttle or cut the internet, Bluetooth mesh keeps working. MeshLive joins the toolkit alongside HF radio and sneakernets: local team voice plus one-to-many audio broadcast with no provider to block. Nothing identifying travels in the clear to strangers, and real names go encrypted only to chosen contacts.
7. Training new hams in net discipline
Before handing a novice a microphone on a repeater, run a practice net on MeshLive audio-only: callsign-style check-ins, directed-net control, relay discipline, prowords. Mistakes cost nothing and touch no spectrum. The mute-to-talk intercom feel transfers directly to push-to-talk habits.
Practical limits hams must respect
MeshLive is beta and unaudited: never make it the sole safety or emergency channel. Amateur radio procedure plus a backup plan stay mandatory.
Range is tens of metres per hop through open air and far less through walls; plan relay phone placement (high, central, screen-off relay volunteers) rather than assuming building-wide coverage.
Video is small and relay chains degrade fast; default to audio-only for anything beyond two hops.
Both ends need Android 10 plus with the same app; iPhones can’t join yet, so mixed groups need loaner handsets.
Bluetooth and Wi-Fi based phone links aren’t amateur spectrum and carry no callsign requirement, but identify per the event plan anyway and keep RF operating separate and legal.
MeshLive Compared with Alternatives
Tool
Bearer
Range model
Voice plus video
Works with no internet
Identity model
License
MeshLive
Bluetooth mesh, phone relays
10 to 30 m per hop, grows with crowd
Yes, plus audio-only broadcast and beta messaging
Yes, fully offline after install
Rotating IDs, QR-verified contacts, E2E encrypted
GPL-3.0-or-later open source
Meshtastic
LoRa 868/915 MHz plus BLE to phone
Kilometres per hop with line of sight
Text plus telemetry, voice only experimental
Yes, needs dedicated LoRa radios
Node IDs, optional encryption
GPL-3.0 open source
Briar
Bluetooth, Wi-Fi, Tor
Bluetooth range or internet via Tor
Messaging plus forums, no calls
Bluetooth yes, full mesh needs Tor
Tor-hidden accounts, in-person verification
GPL-3.0 open source
Bridgefy
Bluetooth mesh
Bluetooth range per hop
Messaging, paid broadcast SDK
Yes after install
Phone-number based
Proprietary freemium
Serval Mesh
Wi-Fi adhoc plus UHF radio
Wi-Fi range per hop
Voice plus messaging
Yes with compatible phones or extenders
Mesh phone numbers
Mixed open source
Handheld FM simplex
Amateur RF 2 m/70 cm
Kilometres, terrain dependent
Voice (and APRS data separately)
Yes, needs license and radios
Callsign on air
Licensed service, open procedure
MeshLive wins when everyone already carries an Android phone and the job is short-range team voice plus live pictures with zero infrastructure. Meshtastic wins for kilometre-scale text telemetry with dedicated radios. FM simplex remains the licensed, long-range, interoperable backbone; MeshLive complements it for logistics that should stay off the amateur frequencies.
Frequently Asked Questions
Does MeshLive need internet, a SIM or an account?
No. Calls, streams and beta messages travel only over Bluetooth between nearby phones. Internet permission exists solely for the optional release check and experimental Wi-Fi Direct APK sharing, never for media.
How far does MeshLive reach in practice?
One Bluetooth hop covers roughly 10 to 30 metres in the open and less through walls. Each relay phone extends the chain, but every hop costs airtime and quality. So plan short chains and prefer audio-only mode.
Can hams use MeshLive for emergency communications?
Yes as a complement, never as the sole safety channel. It suits team talk-around and announcements when cellular data fails, but it is beta and unaudited, so keep amateur radio procedure and backups in place.
Which phones run MeshLive?
Android 10 or newer on both ends. Tested reference phones include Samsung S24, S22, A55 and M31, plus Redmi, ASUS and OPPO models. Other phones may work or hit Bluetooth-stack bugs worth reporting.
How is MeshLive different from Meshtastic?
MeshLive runs on phones you already carry over Bluetooth with voice, video and broadcast; Meshtastic needs dedicated LoRa radios and excels at kilometre-scale text. Many groups will happily run both side by side.
Sources and Further Reading
MeshLive Android repository: https://github.com/MeshLive/meshlive-android
Latest release and APKs: https://github.com/MeshLive/meshlive-android/releases/latest
User guide (install, features, FAQ): https://github.com/MeshLive/meshlive-android/blob/main/docs/user-guide.md
Privacy document: https://github.com/MeshLive/meshlive-android/blob/main/docs/privacy.md
Supported phones: https://github.com/MeshLive/meshlive-android/blob/main/docs/devices.md
Architecture doc: https://github.com/MeshLive/meshlive-android/blob/main/docs/dev/architecture.md
Meshtastic open source LoRa mesh: https://meshtastic.org/
Briar secure messaging: https://briarproject.org/
Obtainium app updater: https://github.com/ImranR98/Obtainium
73 de 9M2PJU. Short-range mesh on the phone, long-range work on the radio.
The post MeshLive: Encrypted Voice Calls, Video Calls and Livestreams Over Bluetooth Mesh with No Internet appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/meshlive-bluetooth-mesh-calls-livestreams-offline-android/

#bluetoothmesh #encryptedvoice #voicecalls #internet #meshlive #quicksummary #project #sim #beta #gatt

#rss #feeds #linksharing #qwibbit

TwinDesk: One Open Source Desk for Mac, Windows, Linux and ...

2026-10-30T18:26:29Z

TwinDesk: One Open Source Desk for Mac, Windows, Linux and Android with No Account and No Cloud

TL;DR: Quick Summary
TwinDesk is a free and open source GPL-3.0 application by Inuka Asith that makes a Mac, a Windows PC, a Linux box and an Android phone work as one desk. One mouse and keyboard drive every screen, files drag across the screen edge, clipboard text and pictures paste on any device, and an Android phone contributes calls, texts, photos, camera, remote control and gamepad input. It works over the same Wi-Fi, a USB cable, Bluetooth, on-demand Wi-Fi Direct, or a phone hotspot, with no account, no cloud relay and no telemetry. Version 2.4.0 covers macOS 13 plus, Windows 10 and 11, Linux on X11 and Wayland, and Android 10 plus. For amateur radio operators this solves the classic multi-device shack problem: run the rig-control PC, the logging laptop and the cluster tablet from one keyboard, answer the phone without leaving the chair, push waterfall screenshots between machines, and keep everything working on Field Day with no network at all.
What Is TwinDesk?
TwinDesk is a free open source device-unification application that shares screens, apps, files, clipboard, cameras, calls, texts, notifications, mouse and keyboard between Mac, Windows, Linux and Android over a direct encrypted link with no account and no cloud.
Install the same app on each computer and phone, pair once with a code, then turn on only the features you want. Push the pointer off the edge of one screen and keep working on the next computer. Open a Windows app on a Mac in its own window. Extend a desktop onto an Android tablet. Answer a phone call from the PC. Copy a screenshot on one machine and paste it on another. Every feature starts switched off on a fresh install and isn’t even loaded until enabled, which keeps idle memory low (about 125 MB in the tray on Windows in v2.3 plus).
The project lives at https://github.com/InukaAsith/TwinDesk with website and live browser demo at https://twin-desk.vercel.app. It holds about 1,400 stars and 109 forks, 269 commits, and ships Electron apps for desktops plus a native Kotlin Jetpack Compose app for Android.
TwinDesk links Mac, Windows, Linux and Android into one desk: extended screens, single apps in their own windows, and the phone as a full peer. Image credit: Inuka Asith, TwinDesk repository, GPL-3.0.
Project Snapshot
Item
Detail
Project
TwinDesk by Inuka Asith
License
GNU GPL-3.0, free and open source
Latest release
2.4.0 (2.3.0 added phone camera, remote, gamepad, picture clipboard, cursor lock)
Platforms
macOS 13 Ventura plus (Apple Silicon and Intel), Windows 10 and 11 (x64 and ARM64), Linux X11 and Wayland (deb, AppImage, Arch, NixOS), Android 10 plus (standard and full APK)
Codebase
Electron on desktops, Swift helper on macOS, C# helper on Windows, C helper on Linux, Kotlin on Android, shared JS protocol core
Network
Same Wi-Fi, USB cable for phones, Bluetooth LE, on-demand Wi-Fi Direct, phone hotspot; TCP 47800 and UDP 47801 on LAN
Privacy model
No account, no cloud relay, no telemetry; devices talk only to paired peers over TLS; only internet request is an optional GitHub update check
Links
GitHub repo, website and demo, releases
Why the Shack Ends Up with Three Keyboards
A typical amateur station grows device by device. A Windows PC runs WSJT-X, FLRig or Ham Radio Deluxe and the soundcard interface. A Linux laptop runs logging, cluster telnet and propagation tools. A phone buzzes with SMS spot alerts, club chat and family calls. A tablet could show a band plan or a logbook but sits unused because there’s no keyboard left for it.
The usual fixes each solve one slice. A hardware KVM switches video but not phones. Shared benefit and Barrier share mouse and keyboard between computers but know nothing about calls, texts or cameras. KDE Connect links one phone to one Linux desktop but not Mac to Windows fleets. Phone Link ties Android to Windows only and routes through a Microsoft account and cloud. Universal Control ties Apple devices to Apple devices only.
TwinDesk aims at the whole desk at once: any combination of computers plus Android as a full peer, over whatever link is available, with each capability behind its own switch.
Computer to Computer: One Desk Across Mac, Windows and Linux
The same app runs on every desktop OS, so a Mac and a PC pair just like two Macs.
One mouse and keyboard
Push the pointer off the configured screen edge and keep working on the other computer as if the two displays were one large surface. The keyboard follows the pointer, including shortcuts with optional Ctrl and Command swapping. Trackpad gestures such as Spaces swipes, pinch to zoom and show desktop carry over. A double tap of the hotkey brings everything home. Edge resistance, pointer speed and which side each peer sits on are configurable per pair.
Version 2.3 added cursor lock: press Scroll Lock (F14 on a Mac), or use the Cursor tab or tray menu, and the pointer stays on its current computer. This matters during contests when a stray edge crossing could dump keystrokes into the wrong log.
Extended display and remote screen
Extend a Mac onto a PC or another Mac with a real virtual display, or mirror it. Alternatively bring the other computer’s screen into a window and control it from there. Streams use low latency H.264 with Fastest and Smooth playback modes plus optional frame generation. On Linux, GNOME Wayland uses the Mutter virtual monitor path, while X11 uses a spare output of the graphics driver (not available on the NVIDIA proprietary driver).
For hams this is the cheapest second monitor available: an old laptop or mini PC becomes the cluster, propagation and log-check display without buying a panel.
The other computer’s apps in windows of their own
Instead of a full remote desktop, TwinDesk can open a single app from the far computer in a window of its own on the local machine, with its own Dock icon or Start menu pin. A Windows-only logger can sit beside native Mac apps. A Mac-only utility can sit beside Windows digital-mode software. On Linux this per-app mode needs an Xorg session.
Camera sharing
Show the other computer’s camera in a window, or expose it as the TwinDesk Camera virtual webcam that Zoom, Teams, Meet or a browser can select. A computer shares only when its owner turns on the allow switch, and a popup on the sharing side names the watcher. Install is one prompt: a Windows 11 camera driver with one administrator approval, v4l2loopback on Linux with one password prompt, OBS Virtual Camera path on macOS.
Files, shared folders and terminal
Drag files across the screen edge and drop them anywhere, with live progress on both sides. Or copy with Ctrl plus Alt plus C and paste with Ctrl plus Alt plus V (Control plus Option plus C and V on macOS), into a folder or straight into the foreground app such as a chat or email. Send from the macOS Share menu or Explorer Send To, or share a folder that appears as a network drive on the other side (Samba based on Linux, Finder mount on Mac). For shell access, ssh twindesk opens a session on the paired computer with no password setup.
Shared clipboard
Copy on one computer and paste on the other, instantly or on switch-over. Text plus pictures up to 10 MB travel, so a screenshot taken with Win plus Shift plus S on the PC pastes directly on the Mac. Password manager entries are skipped by design. A Recent list keeps the last few copies with image previews.
Notifications, sound and Wake-on-LAN
One computer’s notifications appear as native notifications on the other, and audio can flow either way. A sleeping machine wakes via Wake-on-LAN. Every switch is shared by both sides of a pair and remembered, so a contest layout survives a reboot.
More than two devices and TwinDesk Mesh
Pair as many computers and phones as desired. Switch moves the session to another nearby paired device, Take over claims a busy one, Disconnect lets go. TwinDesk Mesh (beta on Windows, Mac and Linux, enabled in the Multi-device tab) connects the whole fleet at once with direct peer links and no hub: one mouse across every screen arranged on a desk map, one clipboard where newest copy wins, one pooled notification stream, and automatic introduction when a new device pairs with any member. Cameras, picture clipboard and multi-target file drag are still being brought into Mesh, phones attach to one computer at a time in Mesh, and Linux Mesh peers need the same Wi-Fi for now.
Phone and Tablet: Android as a Full Peer
Most phone links stop at notifications and file drop. TwinDesk treats Android 10 plus as a full peer that pairs like a computer and adds phone-only capabilities.
Calls and messages
Answer, decline and place the phone’s calls from the computer. An incoming call pops a small window with Answer and Decline plus caller ID. Over Bluetooth the computer acts as the phone’s headset for talk and listen. The Calls page adds a dial pad, contacts and history. The Messages page carries full SMS conversations including pictures, with a sidebar unread count and new-conversation by number.
For an operator mid-pileup, this means no headset swap and no missed call from the net control or family.
Photos and phone files
Browse the phone gallery under Photos, open shots full size, and save one or many to the computer with no cable and no cloud. Phone Files browses whole-device storage with save, upload, rename and delete, plus drag and drop onto the page. Shared folders work both ways: a phone-shared folder opens in Finder, Explorer or the Linux file manager, and a computer-shared folder appears in the phone Files app. The Android share sheet gains Send to computer and Copy to computer entries for any app.
Field photos of antennas, mast work, QSL cards and shack wiring move to the log folder in seconds.
Phone camera as shack webcam
Use the phone back or front camera in a window on the computer, or as the TwinDesk Camera virtual webcam in any meeting app. Switch cameras and rotate from the Camera tab or from the phone Camera page. Phone cameras beat most laptop webcams, which helps for club presentations, remote exam proctoring, livestreams and net check-ins with a view of the bench.
Remote and gamepad
The phone Remote page turns the handset into a touchpad plus keyboard with the keys phone keyboards lack (Esc, Tab, Ctrl, Alt, arrows, media and volume), with landscape rotation. The gamepad layout follows an Xbox controller: buttons can press chosen keys, or Real gamepad mode exposes an actual controller to the PC (Xbox 360 controller on Windows, TwinDesk Gamepad on Linux). A separate Bluetooth gamepad mode pairs the phone in the computer Bluetooth settings with no TwinDesk needed on that host, including Macs.
Tune from the couch, advance slides during a club talk, or drive a media PC from the operating chair.
Media controls and sound
The Overview card shows current phone media with art, Previous, Play or Pause, Next and a seek bar. On Windows and Linux, Play phone sound here turns the computer into the phone Bluetooth speaker. The reverse path plays computer audio on the phone.
Phone screen, single apps and tablet as second screen
Mirror the whole phone screen in a desktop window with mouse, trackpad and keyboard control, or open single Android apps in windows of their own beside desktop apps, pinnable to Dock or Start menu with Enter to send in WhatsApp, Instagram and Telegram. Prop up a tablet and Extend or Mirror the desktop onto it, using touch as mouse or touchpad mode with tap, hold, drag and two-finger scroll.
An old Android tablet therefore becomes a zero-cost logging side screen, cluster board or band scope repeater.
The Android app layout
Phones get bottom tabs, tablets get a full sidebar. Pages include Overview (connection, nearby peers, wireless debugging card), Screen, Cursor, Keyboard, Remote, Camera, Clipboard (per-direction switches), Notifications, Files, Sound, General and About. Outside the app there are share-sheet entries, Quick Settings tiles for Send clipboard and wireless debugging, home-screen shortcuts, and a persistent notification with Exit. Files copied on the computer arrive as a phone notification with Copy and Save actions.
Two APK variants exist: standard for direct download, and full which adds notification access, SMS permissions and the Accessibility service for mouse and keyboard without wireless debugging. Play Protect blocks those capabilities in sideloaded apps, so the easiest path is letting the paired computer install the full APK over wireless debugging from Phone page, Notifications from the phone. Settings and pairing carry over.
Connections: Works Where Field Radio Works
TwinDesk never forces a single transport. It uses the same Wi-Fi when present, moves a phone to USB when plugged in, falls back to Bluetooth LE when Wi-Fi is absent or client-isolated (hotel, cafe, Field Day tent), and raises a fast direct Wi-Fi link only while a screen, camera, file or photo transfer needs it before dropping back to Bluetooth. Press Connect with Wi-Fi Direct to hold the fast link, or let the computer join the phone hotspot with remembered password. The active link and latency stay visible at the top of the window, for example 38 ms over Bluetooth.
Capability
Same Wi-Fi
USB cable
Bluetooth
Plus Wi-Fi Direct or hotspot
Session, clipboard text, notifications
Yes
Yes
Yes
Yes
Mouse and keyboard, remote
Yes
Yes
Yes
Yes
Calls and texts
Yes
Yes
Yes
Yes
Screens, apps, cameras, files, photos, clipboard pictures
Yes
Yes
Auto-moves to Wi-Fi Direct
Yes
Terminal between computers (ssh twindesk)
Yes
No
Over Wi-Fi Direct
Yes
Bluetooth discovery is fenced by two switches: Bluetooth with phones (on by default) and Bluetooth with computers (off until wanted), on Overview and in General. Ports on LAN are TCP 47800 for control plus files and UDP 47801 for discovery.
How Amateur Radio Operators Can Utilise TwinDesk
This is where a generic productivity tool becomes shack infrastructure. The patterns below map directly to contesting, DXing, portable work and EmComm.
1. One keyboard for the contest desk
Put the Windows logger (N1MM Logger Plus, DXLog, WriteLog) on one machine and the digital-mode or spotting machine on the other. Line them up on the TwinDesk desk map so the pointer flows left to right. Type callsigns, exchanges and comments wherever the pointer sits, with no KVM box and no second keyboard crowding the key. Enable cursor lock during a run so a fast mouse flick can’t leak keystrokes into the wrong window in the last ten minutes of a sprint.
2. Digital-mode sidecar without extra monitors
Run WSJT-X or JS8Call with FLRig or Hamlib rigctld on the radio PC, and keep Log4OM, Cloudlog, DX cluster telnet and VOACAP propagation on the second machine. Slide across for a lookup, drag the ADIF export back, paste the worked-before result into the QSO box. Shared clipboard carries frequencies as text (for example 14.074 FT8) and screenshots of decodes for the club chat.
3. Old tablet as dedicated cluster, log-check or satellite panel
Extend the desktop onto a retired Android tablet and park always-on windows there: DXSpider telnet, HamClock or Open Hamclock Backend dashboard, satellite pass list, or a browser-based logger. Touch taps work as clicks. This saves a monitor purchase and keeps RF-quiet low-power displays near the rig instead of another noisy PC.
4. Never miss a call or spot while hands are on the rig
With TwinDesk paired, incoming phone calls and SMS appear on the shack PC as native windows. Answer through the computer headset path, reply to SMS with a full keyboard, and keep the phone charging across the room away from RF pickup. Spot alerts, POTA activation pings and club Telegram messages that arrive on the phone are readable without breaking a run.
Practical tip: mute single noisy apps in the Notifications page before a contest, and optionally hide message text so visitors or stream viewers see only the sender.
5. Move the exact evidence, not a description of it
Waterfall screenshots, antenna analyzer traces, SWR plots, error dialogs and QSL scans travel as picture clipboard up to 10 MB or as edge-dragged files with progress bars. Copy on the measurement laptop, paste into the log notes, email or club forum post on the main PC. Ctrl plus Alt plus V pastes straight into the foreground app instead of forcing a detour through a folder.
6. Clean ADIF, Cabrillo and audio flow
Export ADIF from the log

https://hamradio.my/twindesk-open-source-unified-desk-mac-windows-linux-android/

#opensource #windows #mac #inukaasith #desk #twindesk #cloud #quicksummary #direct #same

#rss #feeds #linksharing #qwibbit

Geomilli: Self-Hosted OpenStreetMap Geospatial Search API with ...

2026-10-30T17:44:30Z

Geomilli: Self-Hosted OpenStreetMap Geospatial Search API with Meilisearch

TL;DR: Quick Summary
Geomilli is an open source Apache-2.0 geospatial platform that converts raw OpenStreetMap extracts into a fast self-hosted place search API. A streaming harvester daemon downloads country PBF files, filters them with Osmium, maps features into 19 canonical classes, enriches them with Wikidata and Wikipedia pageviews, and indexes them into Meilisearch with zero-downtime swaps. A stateless Go API then serves sub-10ms autocomplete, full-text faceted search, reverse geocoding, nearby radius search, bounding-box queries and IATA/ICAO airport lookup. A React MapLibre console called Geomilli Studio provides an API playground, importer control with live progress, and latency telemetry. Default ingest is Belgium in about 60 to 90 seconds via Docker Compose. For amateur radio, it’s a self-hosted gazetteer for SOTA and POTA planning, APRS map labels, repeater and club finders, DXpedition airport routing, and offline EmComm dashboards with no Google API key or usage bill.
What Is Geomilli?
Geomilli is a self-hosted open source geospatial search platform that ingests OpenStreetMap data and serves it through a high-performance Go REST API backed by Meilisearch with sub-10ms autocomplete and multilingual place search.
The repository at https://github.com/obaid22192/geomilli is a monorepo with four parts: geomilli-api (stateless Go REST service), geomilli-harvester (streaming OSM ingestion daemon), geomilli-studio (React 19 plus TypeScript plus MapLibre developer console), and docs (VitePress documentation site published to GitHub Pages). Full docs live at https://obaid22192.github.io/geomilli/. The stack needs Go 1.25 plus, Meilisearch v1.13, Docker Compose v2, and osmium-tool inside the harvester image.
Geomilli Studio map explorer and API playground with live MapLibre map, endpoint tester and one-click cURL generator. Image credit: obaid22192/geomilli repository, Apache-2.0.
Architecture at a Glance
Component
Responsibility
Tech stack
Scaling note
geomilli-api
Autocomplete, search, reverse, nearby, bbox, airports, telemetry
Go 1.25, Chi router, Meilisearch Go SDK
Stateless, scale replicas without limit behind Nginx, Traefik or ALB
geomilli-harvester
Fetch PBF, Osmium filter, transform, Wikidata enrich, index with atomic swap
Go 1.25, osmium-tool, SQLite WAL cache
Single instance only, scale vertically plus HARVESTER_CONCURRENCY
Meilisearch engine
Live places index plus shadow build index
LMDB disk store, local or Meilisearch Cloud
Needs fast NVMe for planet scale
geomilli-studio
Map playground, importer control, telemetry charts
React 19, TypeScript, Tailwind, MapLibre GL
Static frontend pointing at API port 8080
docs
Architecture, pipeline, API reference, studio guide
VitePress with Mermaid plugin
Served from GitHub Pages
Data flow is linear: client apps hit the Go API, the API queries the live Meilisearch index, the harvester builds a shadow index from OSM streams, and a single swap call promotes the shadow to live with zero downtime.
Core Feature 1: Search and Geospatial REST API
Sub-10ms prefix autocomplete
GET /v1/autocomplete?q=Brux&lang=fr returns keystroke suggestions with typo tolerance, localized display names and geo-proximity ranking. A typical hit carries id, display name, class and subclass, country code, per-language names, lat/lng object and a popularity score derived from Wikipedia pageviews. Reported internal timing is around 2 ms.
curl -s "http://localhost:8080/v1/autocomplete?q=Brux&lang=fr"; | jq .
For ham tools this powers callsign-adjacent search boxes: type part of a summit, park or town name and get ranked completions before the operator finishes typing.
Full-text and faceted search
GET /v1/search queries across destinations, categories and attributes with live facet counts. Example from the docs filters restaurants by class, cuisine, star rating and country:
curl -s "http://localhost:8080/v1/search?class=Restaurant&cuisine=italian&star_rating=4&country=BE"; | jq .
The same facet machinery filters places by the 19 canonical taxonomy classes (City, TrainStation, Hotel, Restaurant, Airport, Beach, Stadium and others), which lets a ham app query only summits, parks, stations or airports.
Reverse geocoding
GET /v1/reverse?lat=50.8467&lng=4.3525 resolves coordinates into enclosing admin hierarchy plus nearest landmarks. This is the endpoint APRS dashboards, logger maps and telemetry decoders need: turn a received lat/lng into “near X town, Y district” for display.
curl -s "http://localhost:8080/v1/reverse?lat=50.8467&lng=4.3525"; | jq .
Nearby POIs by radius
GET /v1/nearby?lat=...&lng=...&radius_meters=1500&class=TouristAttraction returns places inside a radius in meters, sortable by distance or popularity. Replace the class with stations, shelters, hospitals or campsites and the same call answers “what is around my activation site” for portable ops.
curl -s "http://localhost:8080/v1/nearby?lat=50.8467&lng=4.3525&radius_meters=1500&class=TouristAttraction"; | jq .
Aviation and airport resolution
GET /v1/airports/BRU looks up by path code, with query filters for IATA, ICAO and country plus limit control. DXpedition and contest travel planning can resolve 3-letter IATA and 4-letter ICAO codes locally instead of scraping airline sites.
curl -s "http://localhost:8080/v1/airports/BRU"; | jq .
curl -s "http://localhost:8080/v1/airports?iata=CRL"; | jq .
curl -s "http://localhost:8080/v1/airports?country=BE&limit=25"; | jq .
Bounding-box queries (/v1/bbox) complete the set for map viewport fetching: send the visible corners, get back only the places inside.
Core Feature 2: Streaming Ingestion Engine
The harvester is a streaming pipeline plus an HTTP/SSE control daemon:
Fetch: streams .osm.pbf country extracts from Geofabrik mirrors for requested countries.
Osmium filter: runs native C++ osmium tags-filter to keep admin boundaries, transport hubs, amenities and tourism features per config/taxonomy.yaml.
Transform: converts geometry to .geojsonseq via osmium export, then emits line-delimited NDJSON batches mapped to 19 canonical classes and filtered through config/languages.yaml.
Enrich: resolves Wikidata IDs, pulls Wikipedia pageviews for popularity scoring, extracts airport IATA/ICAO codes, caches everything in local SQLite WAL cache, and trips a circuit breaker that disables online enrichment after 5 straight network errors.
Index and atomic swap: batches NDJSON into an isolated shadow index named places__build_<timestamp>, configures searchable and filterable attributes, validates, then promotes with a zero-downtime POST /swap-indexes.
Intermediate PBF and NDJSON staging lives on local disk (raw/, staging/), which is why the harvester wants fast local NVMe and can’t share one SQLite file across replicas without SQLITE_BUSY lock contention.
Geomilli Studio: Map Playground plus Ops Console
Studio has three jobs:
API playground: test autocomplete, search, reverse, nearby, bbox and airports against a live MapLibre vector map, with a one-click cURL generator targeting the API on port 8080.
Importer control center: launch and watch country ingestion jobs with real-time Server-Sent Events progress, country dedup locks and daemon logs.
Telemetry dashboard: track QPS, latency percentiles (p50, p95, p99), zero-hit queries and Meilisearch index health with no external APM.
Default ports are API 8080, harvester 8085, Meilisearch 7700, Studio 3333 (or 3000). Useful Makefile targets are make up, make down, make logs, make test, make build, make docs, make docs-build and make up-cloud.
60-Second Quickstart
Default config ingests Belgium (about 70 MB, ready in 60 to 90 seconds):
git clone https://github.com/geomilli/geomilli.git
cd geomilli
cp .env.example .env
docker compose up -d --build
Then open the API at http://localhost:8080, harvester at http://localhost:8085, Meilisearch at http://localhost:7700, Studio at http://localhost:3333, and docs at https://obaid22192.github.io/geomilli/. Set COUNTRIES in .env for more extracts, raise HARVESTER_CONCURRENCY (default 2) for parallel country jobs on a strong host, or point MEILI_HOST and MEILI_API_KEY at Meilisearch Cloud and start with the cloud compose override.
How Amateur Radio Operators Can Use Geomilli
1. SOTA, POTA and WWFF activation planning
Portable awards live or die on place data: summit names, park boundaries, parking, trailheads, nearby towns for access directions. Ingest the target country extract, then use autocomplete for summit and park lookup, nearby radius search for parking and campsites around the reference coordinates, and reverse geocoding to confirm the parking pin sits inside the right admin area. A club activation page can offer “type three letters, pick the peak” with no Google Places bill.
2. Self-hosted gazetteer for loggers and cluster maps
Logging tools constantly need “QTH name to coordinates” and back. Geomilli gives both directions locally: autocomplete and search for forward lookup, reverse for coordinate to place name. Multilingual names (names_by_language) handle border areas and DX QTHs where the English, French and Dutch names differ. Popularity scoring keeps big cities above same-named hamlets in results.
3. APRS and telemetry map labels
An APRS-IS dashboard receiving position packets can call /v1/reverse per fix and render “9M2PJU-9 near Rawang, Selangor” instead of bare digits. Nearby search adds context layers: nearest town, station or shelter within X meters. Because the API is stateless and horizontally scalable, a busy net night just means more API replicas behind Nginx.
4. Repeater, club and net finder backend
Ingest a country, filter to relevant classes, and build a “find my nearest” page: nearby search around the browser geolocation sorted by distance, faceted search by class for clubs and landmarks used as net venues. Pair with your own repeater table for RF data and let Geomilli supply the place layer (town names, admin areas, coordinates) so the directory stays consistent.
5. DXpedition and contest travel routing
The airports endpoint resolves IATA and ICAO codes offline from your own stack. Plan the “fly to X, drive to the island” leg with country=BE&limit=25 style enumeration, confirm the destination airport coordinates, then radius-search accommodation and transport hubs around it. No airline site scraping, no rate limits.
6. EmComm and offline field dashboards
Emergency communications teams need hospitals, shelters, stations, stadiums and admin boundaries when the internet is degraded. A pre-ingested Geomilli instance on a laptop or Raspberry Pi class host serves autocomplete, reverse and nearby with no external dependency after ingest. The harvester circuit breaker already tolerates flaky links by disabling enrichment after repeated failures while cached data keeps serving. Keep the index built before deployment day, then run the API plus Meilisearch locally at the command post.
7. Why self-host instead of Nominatim, Photon or Google
Option
Hosting
Speed profile
Data control
Cost and limits
Geomilli
Self-hosted Docker stack, Apache-2.0
Sub-10ms autocomplete via Meilisearch
Choose countries, taxonomy, languages, popularity model
No per-query bill, hardware cost only
Nominatim
Self-host or public API
Correct but heavier, Postgres based
Full OSM planet logic, complex setup
Public API has strict usage policy
Photon
Self-host Java plus Elasticsearch
Fast typeahead over OSM
Good defaults, less tunable ranking
JVM plus ES footprint
Pelias
Self-host microservices plus ES
Strong geocoder, many moving parts
Very configurable, complex ops
Heavy multi-container ops burden
Google Places/Geocoding
Cloud only
Excellent
None, terms can restrict storing results
Per-request billing, key management
Geomilli fits hams who want a small number of countries, typo-tolerant typeahead, airport codes and a MapLibre playground in one compose file, without running a full planet Nominatim database.
Practical Tips and Limits
Start with one small country (defaults to BE) and confirm the pipeline before ingesting large neighbors. Malaysia (MY) and neighbors follow the same COUNTRIES pattern via Geofabrik paths.
Give the harvester fast local disk and enough cores; it is single-instance by design, so scale up the host and raise HARVESTER_CONCURRENCY rather than adding harvester replicas.
Scale the API freely: it holds no local state, so add replicas behind any reverse proxy.
Watch telemetry for zero-hit queries: they reveal missing taxonomy entries or language gaps worth adding to taxonomy.yaml and languages.yaml.
Respect licenses: code is Apache-2.0, OpenStreetMap data stays under ODbL with attribution.
The repo is young (7 commits, 2 stars at write time), so pin a release commit, test upgrades on a spare host, and keep Geofabrik plus Wikidata outage fallbacks in mind for enrichment.
Frequently Asked Questions
What does Geomilli do in one sentence?
Geomilli converts OpenStreetMap country extracts into a fast self-hosted search API with autocomplete, reverse geocoding, nearby search and airport lookup backed by Meilisearch.
How fast is Geomilli search?
Autocomplete is designed for sub-10ms responses with typo tolerance and popularity ranking. The API itself is stateless Go, so extra replicas behind a proxy handle contest or net-night load.
What hardware runs a country ingest?
Any Docker host with decent CPU and fast local disk works. Belgium at about 70 MB finishes in roughly a minute. Bigger countries want NVMe staging and higher harvester concurrency on a single instance.
Can Geomilli work offline at a field station?
Yes, after ingest. Build the index with connectivity, then run the API plus Meilisearch locally at the site for place search, reverse labels and nearby POIs with no external calls.
How can hams use Geomilli concretely?
Use it for SOTA and POTA site lookup, APRS coordinate-to-place labels, repeater and club finder place layers, DXpedition airport routing, and offline EmComm dashboards for shelters and stations.
Sources and Further Reading
Geomilli repository by obaid22192: https://github.com/obaid22192/geomilli
Geomilli documentation site: https://obaid22192.github.io/geomilli/
Meilisearch search engine: https://www.meilisearch.com
OpenStreetMap copyright and license: https://www.openstreetmap.org/copyright
Geofabrik OSM extracts: https://download.geofabrik.de/
Osmium tool manual: https://osmcode.org/osmium-tool/
MapLibre GL JS: https://maplibre.org/
Nominatim usage policy: https://operations.osmfoundation.org/policies/nominatim/
Summits on the Air programme: https://www.sota.org.uk/
Parks on the Air programme: https://parksontheair.com/
73 de 9M2PJU. Host your own map intelligence, then take it portable.
The post Geomilli: Self-Hosted OpenStreetMap Geospatial Search API with Meilisearch appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/geomilli-openstreetmap-geospatial-search-api-meilisearch/

#api #geospatialsearch #meilisearch #geomilli #dockercompose #quicksummary #googleapi #react #map #studio

#rss #feeds #linksharing #qwibbit

Stream Deck for Ham Radio: N4MI’s One-Press Shack Automation ...

2026-10-30T17:03:50Z

Stream Deck for Ham Radio: N4MI’s One-Press Shack Automation with Mode Launchers, Propagation Dashboard, and IC-7610 Control

TL;DR: Quick Summary
Stream Deck for Ham Radio by Dan Marshall N4MI is a free open source project that turns an Elgato Stream Deck into a one-press control surface for a Windows ham shack. Tap FT8 and the full digital stack (ACLog, WSJT-X, JTAlert, rotator software) launches in order with correct delays; tap CW or SSB and the matching stack comes up instead. Other buttons park everything cleanly, swing a Yaesu rotator to preset DX headings via PSTRotatorAz UDP, open a locally hosted propagation dashboard with live solar data and three-level tower storm alerts, drive an Icom IC-7610 over its Ethernet port without touching the logger’s USB COM port, aggregate ham blogs and podcasts in HamShackFeed, and catalog radio manuals in ManualShelf. Everything is PowerShell, Python, and HTML you can read and adapt, MIT licensed at github.com/N4MI73/streamdeck-hamradio, documented across a nine-part blog series at n4mi.tech. It is built for one specific station, so expect to edit paths, ports, and callsigns, ideally with an AI assistant helping.
What Is Stream Deck for Ham Radio?
Stream Deck for Ham Radio is a free open source collection of PowerShell scripts, Python web servers, and HTML dashboards that converts an Elgato Stream Deck into a purpose-built ham shack controller, launching complete mode software stacks, steering an antenna rotator, serving live propagation data, aggregating ham content, cataloging manuals, and operating an Icom IC-7610 over LAN from labeled hardware buttons.
The project lives at github.com/N4MI73/streamdeck-hamradio under the MIT license, built by Dan Marshall N4MI of Grovetown, Georgia (EM83) with Claude AI assisting on the programming while N4MI tested every behavior against real hardware. With 92 commits it’s an actively maintained personal system published as a working example rather than a turnkey product. A companion blog series at n4mi.tech walks through each part with screenshots and setup detail across nine installments, from mode launchers to IC-7610 rig control.
N4MI’s Stream Deck main screen with one-press shack controls, including the RADIO folder that opens IC-7610 band and mode buttons. Image via the project announcement on Reddit.
Why a Stream Deck Belongs in the Shack
Every HF operator knows the startup ritual. For FT8 you open the logger, then WSJT-X, then JTAlert, in that order, wait for each to settle, check the rotator software, confirm the right COM port survived, and only then call CQ. For CW it’s a different stack with a Morse decoder; for SSB something else again. Shutdown reverses the dance, and forgetting a step leaves a zombie process holding a COM port that breaks tomorrow’s session. Multiply by daily operation and the overhead is real.
The deeper problem is contention. On N4MI’s desk the IC-7610’s USB CI-V port already has an owner: N3FJP ACLog’s rig interface, or WSJT-X during digital sessions, and only one program can hold a COM port at a time. Anything new that wants to talk to the radio has to fight the logger for it. The Stream Deck project answers both problems with a hardware control layer: one press runs an ordered, delayed, reminder-equipped startup for each mode, and radio control bypasses USB entirely by using the IC-7610’s Ethernet port, so the logger never notices except that its frequency display follows along.
A $150 content-creator macro pad becomes, in effect, a custom shack controller of the kind that used to require dedicated hardware or a fully scripted PC. The buttons are physical, labeled with a consistent icon set, and safe to hit mid-QSO without looking at a screen.
One-Press Mode Launchers: FT8, CW, and SSB Stacks
The heart of the system is scripts/HamRadioLauncher.ps1, invoked from Stream Deck buttons through BarRaider’s free Advanced Launcher plugin with arguments like -mode ft8, -mode cw, or -mode ssb.
Each mode launches its complete application stack in dependency order with tuned delays between steps, so WSJT-X never starts before the logger owns the COM port. Popup reminders prompt for ACLog configuration details that humans forget, and steps needing administrator rights elevate themselves automatically. A companion CloseHamRadio.ps1 shuts everything down in correct dependency order and reminds the operator to return the rotator to North, which is exactly the step you skip at midnight after a long session.
Configuration is deliberately plain text at the top of the scripts: executable paths for ACLog, WSJT-X, JTAlert, MRP40 Morse decoder, and PSTRotatorAz, plus COM port references (COM 6 for the IC-7610, COM 10 for the WinKeyer on N4MI’s desk), UDP ports, Chrome profile names, and launch delays like $shortDelay and $jtalertDelay for slower PCs. You copy the tree to C:\Ham Scripts\, edit the paths in Notepad, and bind each button to pwsh.exe or powershell.exe with -ExecutionPolicy Bypass -File arguments. One caution from the rig-control notes generalizes: PowerShell 7 from the Microsoft Store or winget lives in a protected folder the Stream Deck software can’t reach, so buttons fail silently; use the MSI install in C:\Program Files\PowerShell\7\ if you want pwsh.
Propagation Dashboard and Three-Level Storm Alerts
dashboard/ hosts a locally served propagation portal: dashboard_server.py (a small Python server that also proxies DX cluster data) plus N4MI_PropagationDashboard.html, launched and stopped from Stream Deck buttons via PID-file scripts. One press brings up live solar indices, band conditions, DX cluster spots, and propagation maps in a personal operating portal.
The standout feature is the storm alert system, built for a tower owner. A three-level banner (CAUTION, WARNING, CRITICAL) fuses WeatherFlow Tempest lightning data with NOAA National Weather Service alerts and warns when it’s time to lower the tower. The Tempest token lives in a Windows user environment variable named TEMPEST_TOKEN, never in the script files, and the server prints at startup whether it found one; without a token the alerts run on NWS data alone. Since October 2026 the dashboard server answers only the dashboard page and its /api/ endpoints, so stray files in its folder aren’t served to the network. Operators who want the portal always on rather than tied to the shack PC can run the Docker edition in the sibling n4mi-shack-server project, which reads the same token from an environment variable.
This same always-on logic is what later produced PropMon, the NAS-hosted propagation JSON backend feeding N4MI’s ESP32 desk instruments. PropMon started life in this repo’s propmon/ folder and moved to n4mi-shack-server on 19 October 2026. If you read our earlier PropMon guide, that’s the same software grown up: the dashboard taught the household that anything worth glancing at mustn’t depend on a PC that gets switched off.
Rotator Control: Compass Buttons and Preset DX Headings
scripts/RotatorAzimuth.ps1 and RotatorStop.ps1 send azimuth commands straight to PSTRotatorAz over UDP (enable Communication, UDP Control on port 12000 in PSTRotatorAz). Stream Deck buttons hold compass directions plus preset headings for key DX regions from N4MI’s EM83 grid in central Georgia: Europe at 50 degrees, West Africa at 80, South America at 145, the Pacific at 290, and Asia at 342. Any -az value from 0 to 360 works, and a dedicated stop button halts rotation immediately.
Operators outside Georgia must compute their own great-circle headings with ACLog, PSTRotatorAz’s DXCC lookup, or an online calculator, then update the -az arguments on their buttons. The hardware expectation is a Yaesu rotator with the ERC Mini USB interface, and the UDP port is one $udpPort variable away from matching whatever PSTRotatorAz is configured to use.
Web Launcher: The Contest-Grade Browser Setup
scripts/HamRadioChrome.ps1 opens the operator’s ham radio tabs (QRZ, LoTW, Club Log, PSKReporter, DX cluster) in a dedicated Chrome profile, separate from everyday browsing. A dedicated profile matters more than it sounds: it keeps cluster logins, LoTW certificates, and dark-theme dashboard tabs out of the way of normal web use, and one button rebuilds the whole operating browser after a reboot. The default profile name is Profile 1, a single variable at the top of the script.
HamShackFeed and Pro: A Personal Ham Content Aggregator
HamShackFeed solves feed fatigue: ham blogs, podcasts, and YouTube channels scattered across dozens of sites, each checked never. The original HamShackFeed/HamShackFeed.html is a single self-contained file aggregating everything into one searchable dashboard with favorites, read tracking, and a built-in podcast player, fetching through the rss2json.com proxy. It needs a free rss2json API key pasted into RSS2JSON_KEY (without one the proxy throttles to one request per 10 seconds and adding sources fails), which gives 10,000 requests per day on the free tier.
HamShackFeed Pro in hamshackfeedpro/ removes that dependency with a Flask server, direct RSS parsing, SQLite persistence, background refresh every 30 minutes, and LAN-wide access on port 8074, so read state follows you between shack PC, laptop, and phone. A migration script imports custom sources from the HTML version’s localStorage, and YouTube channels should be added by full channel URL rather than handle. The tradeoff table from the repo says it best:
Feature
HamShackFeed (single file)
HamShackFeed Pro (server)
RSS proxy API key
Required (rss2json, free tier)
Not needed, direct fetching
Read/source persistence
Browser localStorage
SQLite database
Cross-device access
No, single browser
Yes, across the local network
Background refresh
No, refresh on open
Every 30 minutes automatically
Hosting
Open the file
Python server on port 8074
Pro runs on the shack PC or as an always-on Docker container in n4mi-shack-server, with a Windows Firewall rule opening TCP 8074 for other devices.
ManualShelf: The Radio Manual Catalog
Every shack accumulates PDFs: rig manuals, amplifier guides, antenna analyzer references, screenshots of menu settings that took an hour to perfect. manualshelf/ catalogs them in a Flask app on port 8075 with auto-generated thumbnails from PDF first pages via PyMuPDF, tagging, full-text search across indexed PDF contents (with a purple PDF TEXT badge on text matches), a bulk tag editor, and an update flow where re-scanning a folder flags changed files with an amber UPDATE badge while preserving titles and tags.
The cross-computer design is clever: each instance writes a manifest to a shared NAS exchange folder every 5 minutes, and the other machine lists those remote manuals under a Remote Manuals filter with a note that the file must be fetched via LocalSend or the NAS share. Since October 2026 ManualShelf binds to the local computer only, because its API can open desktop files and has no login; set MANUALSHELF_HOST=0.0.0.0 to restore LAN access at your own risk. Supported formats span PDF, images, Office documents, text, markdown, and HTML, while executables and scripts are refused outright.
IC-7610 Rig Control over LAN: The Flagship Addition
The October 2026 addition, documented in rigcontrol/ and in the latest n4mi.tech post, is the technically deepest part of the project: full band, mode, filter, tuning-step, split, preamp, NR, and NB control of the Icom IC-7610 from Stream Deck buttons, sent over the radio’s Ethernet port instead of USB.
The architecture is one PowerShell script plus a hidden helper: each button runs powershell.exe rig.ps1 <action>, the script talks to rigplane (open source Python library for Icom’s LAN protocol) serving a rigctld-compatible endpoint on 127.0.0.1 port 4532, and rigplane logs into the radio over UDP ports 50001 to 50003 as a network user. The first press of the day starts the helper and logs in over about 5 to 10 seconds; every press after that is near instant because the session stays up. Every press begins by reading the radio’s frequency, so a dead helper is restarted automatically (self-healing), a failed press plays the Windows error sound, and a rolling rig.log records the reason. The helper binds to localhost only, since it can key the transmitter, and the radio password is stored in rig_cred.xml encrypted with Windows DPAPI so only the same user on the same PC can read it.
The button set mirrors real operating: ten band buttons (160 through 6 m) that save your current frequency, mode, and filter into the departing band’s stack register before recalling the new band’s, exactly like the front-panel keys; SSB with automatic LSB below 10 MHz and USB above; CW with narrow FIL3; USB-D1 data mode for FT8 and FT4; a FILTER cycler; plus and minus 100 kHz steps that stop at US band edges instead of leaving the band; one-press SPLIT UP 5 for SSB and UP 1 for CW pileups (copy MAIN to SUB, offset, split on, refused with an error tone if transmit would land out of band); PREAMP cycling; NR and NB toggles; and STOP to release the LAN session. Split safety is enforced globally: band, mode, and tuning buttons drop split first so transmit never strands itself on the wrong band or mode. Sixty metres is absent because the 7610 has no band-stacking register for it, and remote power-on was tried and dropped as unreliable.
Real-hardware testing taught lessons no simulator would: 6 m band codes are BCD so the band reads 0x10 not 0x0A, a mode change can silently restore that mode’s last filter so the command is sent twice and verified by readback, computer-issued band changes don’t save your departing spot so the script writes the register itself, and two rigplane 2.11.1 commands stall two seconds waiting for replies the radio never sends, so the script uses equivalent CI-V commands that return promptly (the project pins rigplane 2.11.1 for exactly this reason). Setup takes about fifteen minutes: enable Network Control on the radio and power-cycle it, set a network user and password with a DHCP reservation, copy the folder, install rigplane in a venv, set $RadioIP, run rig.ps1 setup, and bind buttons. Owners of other LAN-capable Icoms like the IC-705 or IC-9700 can adapt it with different CI-V addresses and band codes, testing every button before trusting it on the air.
Custom Icons and Adapting It to Your Station
The icons/ folders hold ready-made 144×144 PNG buttons in one consistent design language, including 26 rig-control icons, with Pillow scripts (make_icons.py) that regenerate them so you can change labels or colors in code rather than by hand. The author’s layout nests a RADIO folder on the main page holding ten band buttons, STOP, and a MODES folder with the rest.
The repo is explicit that this is a starting point, not a product: paths, COM ports, UDP ports, Chrome profiles, callsign and grid, Tempest tokens, DX headings, launch delays, band edges (US allocations), filter choices, split offsets, and the radio IP all encode one Georgia station. The good news, as the README notes, is that AI assistants adapt this kind of well-organized script collection far faster than writing from scratch, and N4MI’s own series was built exactly that way, with the human testing on real hardware and the AI doing the programming.
Stream Deck Shack Control Versus the Alternatives
Approach
Cost
Operating feel
Rig integration
Shack-wide scope
N4MI Stream Deck system
Stream Deck hardware plus free software, MIT
Physical labeled buttons, one press per task
IC-7610 over LAN, no COM fight; PSTRotatorAz over UDP
Launchers, dashboard, rotator, feeds, manuals, rig in one repo
Touch Portal / Macro Deck phone app
Free or low cost app
Touchscreen, no tactile keys
Via plugins and scripts, DIY
Comparable if you build it all
Dedicated controllers (RC-28 knob, band decoders)
Hundreds of dollars
Excellent single-purpose feel
Native, purpose built
One device per job, no launcher logic
Pure software (SDR console, logger macros)
Free
Mouse and keyboard only
Depends on CAT sharing
Limited to each app’s own window
Voice or fully automated control
Varies
Hands free but brittle
Script dependent
High setup cost
The Stream Deck’s edge is tactile one-press reliability across the whole shack rather than depth in any single function: no single button here beats a dedicated controller. But no dedicated controller launches your logger stack, checks the solar wind, parks the rotator, and files your manuals.
Frequently Asked Q

https://hamradio.my/n4mi-streamdeck-ham-radio-shack-automation/

#streamdeck #hamradio #modelaunchers #propagationdashboard #elgatostream #shack #danmarshall #control #summarystream #deckbelongs

#rss #feeds #linksharing #qwibbit

NightDX: Unattended Overnight Shortwave and Medium Wave Listening ...

2026-10-30T16:23:42Z

NightDX: Unattended Overnight Shortwave and Medium Wave Listening with the AirSpy HF+

TL;DR: Quick Summary
NightDX is a free open source Python application by Marco Zennaro that turns an AirSpy HF+ software defined radio into an unattended overnight broadcast DX station. Before bed you arm a night mission, the software scans every band to measure what is really audible, predicts which scheduled broadcasters should propagate to your location, records the most promising stations through the night with a discovery mode that prefers stations you have never logged, then classifies each recording as speech, music, or noise with an optional on-device AI model and transcribes speech with Whisper. In the morning a browser report presents every catch with an audio player, a one-click reception report generator with SINPO and program details that files as a Gmail draft with the recording attached, and a day-by-day log with QSL card tracking. Schedules come from EiBi and AOKI plus broadcaster HFCC technical data downloaded automatically, propagation uses a compact VOACAP-style HF model and an ITU-R medium wave model with live NOAA solar data and self-calibration from your own scans, and the whole interface is a keyboard-driven 1990s style terminal UI. It runs on macOS, Windows 10/11, and Linux, installs with one uv command, and is MIT licensed at version 0.9.10.
What Is NightDX?
NightDX is a free open source unattended shortwave and medium wave broadcast listening station built around the AirSpy HF+ receiver, combining automatic schedule downloads, physics-based propagation prediction, real band-scan measurements, overnight scheduled recording, optional AI audio classification and transcription, reception report generation, and QSL card logging in a single menu-driven terminal application.
The idea is simple and aimed at a real shortwave listener problem. The best broadcast DX happens while you sleep: transatlantic medium wave openings peak in the small hours, Asian and Pacific shortwave targets line up with greyline at dawn, and nobody wants to sit at the radio at 3 AM pressing record. NightDX closes that gap. You spend five minutes in the evening checking the plan, arm the mission, go to bed, and wake up to a morning report that tells you what was on, what it sounded like, and which stations are worth a reception report. Every recording is already named after its most likely station, already measured, already auditioned by the AI, and one click away from a properly formatted report email.
The project lives at github.com/marcozennaro/nightdx under the MIT license, written in Python on the Textual terminal UI framework. At the time of writing it is at version 0.9.10, with 19 commits since the 0.9.0 release candidate on 19 October 2026, and development is visibly active with several releases per week driven by real on-air use from the author’s station in Trieste, Italy, plus early testing reports from Windows users with real HF+ hardware.
NightDX in action: the 1990s style terminal interface with the night mission dialog armed over the main screen, showing ranked countries, radio status, schedules, and live space weather. Image via the project announcement on Reddit.
Why Unattended Listening Matters for SWL and MW DX
Shortwave and medium wave broadcast DX has always been a scheduling problem as much as a radio problem, and three facts make unattended recording almost mandatory for serious results.
First, propagation dictates the timetable, not the listener. The D layer absorbs medium wave and lower HF signals during daylight, so transatlantic medium wave DX into Europe opens roughly between local midnight and sunrise, while Asiatic shortwave into Europe peaks around dawn greyline. A listener in Malaysia chasing European evening broadcasts faces the mirror image: the opening happens on somebody else’s night. Either way the best signals arrive when a human operator is least alert.
Second, broadcast schedules are dense and volatile. On a typical evening the 49, 41, 31, and 25 metre bands carry thousands of scheduled transmissions from the EiBi and AOKI lists, many overlapping on shared channels, with seasonal A/B schedule changes twice a year plus unannounced frequency swaps. No human can monitor more than one or two frequencies at once, so manual listening samples a tiny fraction of what is actually audible.
Third, verification takes longer than listening. A proper reception report needs the station name, frequency, date and UTC time, SINPO code, program details sufficient to prove you heard them, and often an audio clip. Writing that up from a live listening session means pausing to note details while the signal fades. A recorded file with metadata, an AI transcript, and a signal measurement turns a frantic scribble into a calm morning task.
NightDX attacks all three: it predicts using schedules plus propagation, measures using real band scans, records the best candidates automatically, and packages the evidence for reporting. The closest mainstream equivalents are KiwiSDR scheduled recordings and general SDR software timers, but those record blind. NightDX records with a reason attached to every file.
The Night Mission Workflow: Arm Before Bed, Read the Report at Breakfast
The signature concept is the night mission, and it is worth understanding as a workflow before diving into components.
In the evening you open NightDX and press F6. A dialog asks for the mission window, typically something like 22:00 to 06:30, plus optional targets if you’re chasing specific stations or countries. The F4 plan preview shows what the software intends to record and why, ranked by predicted audibility. If something looks wrong you press I on any entry for the full reasoning: schedule source, transmitter power and beam from HFCC data, predicted field strength, live solar indices, the learned self-calibration correction from your own past scans, and the resulting score.
You arm the mission and walk away. NightDX keeps the computer awake, steps through the plan frequency by frequency, records each slot as FLAC with full station metadata, and moves on. The Windows build runs the mission hidden in the background and keeps going even if you close the main window. A discovery mode biases the scheduler toward stations and countries missing from your log, so the system actively works through your DXCC-style wanted list instead of re-recording the same strong routine stations every night.
In the morning you press F8 or open the HTML report in a browser. Every recording appears with the AI verdict (speech, music, or noise), an embedded audio player, the measured signal against the prediction, the other stations scheduled on the same channel, carrier offset measurements, and a Send report button that remembers whether you already wrote to that station. You listen, confirm the identification with your own ears, click through the genuine catches, and file the rest. That’s the whole loop: five minutes at night, fifteen minutes in the morning, and a logbook that fills itself.
Schedules: EiBi, AOKI, HFCC Technical Data, MWLIST, and Your Own List
Prediction is only as good as its input data, and NightDX layers five schedule sources.
EiBi is the backbone: Eike Bierwirth’s long-running shortwave broadcast schedule, the standard reference the whole SWL community uses, covering international broadcasters with frequency, time, language, target area, and site. The AOKI list from the Nagoya DX Circle complements it with additional monitoring-based entries, catching stations and seasonal changes the main list misses. Both download automatically with nightdx update.
HFCC data is the differentiator added in version 0.9.7. The High Frequency Coordination Conference publishes the broadcasters’ own coordinated technical schedules: real transmitter power, beam azimuth, and antenna type for roughly two thirds of shortwave broadcasts. NightDX matches about 3,000 of 4,800 shortwave entries against EiBi by frequency, time, and site, and folds the real power and beam into the prediction. Tested on roughly 1,000 recordings from Trieste, this single change lifted ranking quality from 0.64 to 0.68 AUC and the correlation of predicted with measured signal from 0.37 to 0.48. HFCC lines that EiBi doesn’t confirm are deliberately excluded, on the sound principle that not everything registered actually goes on the air.
MWLIST covers medium wave properly. EiBi and AOKI handle international MW broadcasters, but the full MW band needs the community MWLIST/FMSCAN database. Registration at fmlist.org is free: log in at fmscan.org, set your location, download the Perseus/ELAD/Winradio user list in CSV tab-separated format, drop it in the mwlist folder inside the data directory, and press Ctrl+U. Note the MWLIST data is licensed for personal non-commercial use.
Your own stations joined in 0.9.6 via my_stations.csv, opened from Settings and edited in any text editor with format guidance at the top of the file. One broadcast per line with frequency, UTC window, days, station, and optional country, language, target, coordinates, and power. Only frequency, time, and name are required; without coordinates the transmitter is placed in the named country. NightDX reloads the file on save, reports unreadable lines, and treats your entries as first-class citizens in predictions, scans, missions, and reports, replacing any schedule entry with the same frequency and name. This is the correct home for pirates, clandestines, local low-power stations, and corrections the big lists get wrong.
Propagation: Compact VOACAP-Style HF, ITU-R Medium Wave, Live Solar Data, Self-Calibration
With schedules loaded, NightDX needs to answer the core question: which of these 4,800 broadcasts will actually be audible at my station tonight? It uses real physics, not heuristics.
On shortwave the model walks the great-circle path in F2 hops, derives critical frequency foF2 from local time, solar flux, and Kp index, computes maximum usable frequency from hop geometry, applies George and Bradley D-layer absorption, and adds ITU-R P.372 atmospheric noise. On medium wave it switches to an ITU-R P.368 ground-wave fit plus the ITU-R P.1147 sky-wave formula, which is the right tool for the job since MW night propagation is a different physical regime from HF skip.
Live space weather comes from NOAA SWPC: solar flux and Kp feed the model on every run, so a geomagnetic storm automatically demotes high-latitude paths in the evening plan. Pressing F3 shows the most likely countries right now, ranked with measured against predicted values side by side.
The most interesting part is self-calibration. After every band scan, NightDX learns the gap between prediction and measurement. Since 0.9.8 the learned correction is split into a receiver-wide component (your antenna, your local noise floor, the model’s systematic bias, measured around minus 33 dB at the author’s station) applied to every station, plus a bounded per-station deviation held within plus or minus 10 dB on shortwave. The earlier design applied only per-station corrections, which meant unscanned stations with no correction jumped ahead of calibrated ones and a few wild values up to plus or minus 40 dB poisoned predictions. The split design tested on about 800 shortwave recordings improved AUC from 0.68 to 0.72 and rank correlation from 0.38 to 0.48. Every scan measurement used for learning is appended to scan_pairs.jsonl in the data folder, capped at 20 MB, so future calibration changes can be validated against real history. Pressing I (why this score) exposes both correction components for any entry, which is exactly the transparency a technical hobbyist wants.
Band Scan: Measuring What Is Really Audible
Models predict, but the band scan measures. Press Ctrl+R and NightDX snapshots every band through the HF+ in 768 kHz chunks of about half a second each, completing a full sweep in roughly 10 seconds, then compares each scheduled carrier against the local noise floor.
The main screen presents the payoff directly: countries ranked by measured Heard dB next to the propagation prediction. When measurement and model agree, confidence is high. When they disagree, the disagreement is data: either conditions differ from the solar indices, or the schedule is wrong, and the self-calibration machinery records the lesson.
Two practical details matter. The scan uses libairspyhf directly, so no other SDR program can hold the radio at the same time; close SDR#, SDR++, or HDSDR first. And the receiver is tuned 10 kHz above the station by default (adjustable under Settings, Receiver, Tune offset), with NightDX shifting the signal back in software, which keeps the wanted signal clear of the small DC spike at the center of the receiver spectrum. This isn’t a frequency correction, just clean engineering, and the frequency you hear and record is exactly the one you chose.
Recording, Listen Live, and the Retro Terminal Interface
Recording is AM demodulation straight from the IQ stream, saved as FLAC with full station metadata embedded. The interface tradecraft around it is thoughtful.
Listen live (key L, added in 0.9.9) plays the highlighted frequency through the computer speakers without recording, complete with a signal meter and the stations scheduled there. Tune with minus/plus keys, Page Up/Page Down for 5 kHz steps, or type a frequency directly, and the Record button converts a live session into a normal recording at any moment. The 0.9.10 release fixed the station information panel to a constant height so the tuning buttons stop jumping around while stepping through the band.
The essential keyboard map is worth learning: F2 stations on the air now, F3 most likely countries, F4 tonight’s plan preview, F5 or Enter record the highlighted station, F6 night mission, F7 listening log, F8 report, F9 settings, I why this score, L listen live, X skip the station in night missions, Ctrl+R scan now. F1 shows the full key list. Any list sorts by clicking a column title, click again to reverse, third click restores the default order.
Skipping (key X, also 0.9.9) deserves a mention because it solves a real annoyance: some schedule entries are permanently wrong for your location, or simply uninteresting. Skipping excludes a station on that frequency or on all frequencies from night missions, lists are marked accordingly, and Night, Skipped stations restores them with Enter.
The aesthetic is deliberate: a menu-driven text interface that looks like 1992, built on Textual. No browser dashboard, no Electron shell, no GPU required. It runs over SSH, sips resources during overnight missions, and every function also exists on the command line (nightdx update, nightdx best -n 20, nightdx record 9420, nightdx plan, nightdx night --now --end 06:30, nightdx report), which makes cron and Windows Task Scheduler automation straightforward.
AI Audio Verdicts and Whisper Transcripts (Optional)
The optional ai extra, about 2 GB plus models on first use, adds two machine listening stages to every recording.
First, the MIT AST AudioSet transformer classifies 10-second windows as speech, music, or noise, with the label set cleaned in 0.9.1 to radio-relevant tags only so noisy shortwave audio no longer comes back tagged Vehicle or Insect. This triage is the real value: a night mission can produce dozens of files, and knowing which ones contain actual program audio versus empty-channel hiss decides where your morning listening time goes.
Second, OpenAI Whisper Small transcribes and translates speech for the reception report. Two design choices stand out: weak signals are left blank rather than guessed, which is the honest behavior for evidence-grade reporting, and the detected language is compared against the scheduled language as an identification check (more on that below).
Both models run locally on your machine. Nothing is uploaded to any cloud service, which matters for a hobby that already hands enough data to third parties.
Honest Identifications: Candidates, Carrier Fingerprints, and Co-Channel Rivals
The standout engineering honesty in NightDX, developed across versions 0.9.3 through 0.9.5, is its treatment of station names as candidates, not identifications.
On a crowded channel, and medium wave above all, the scheduled station may not be what you actually hear. So every recording carries its evidence: the language Whisper heard versus th

https://hamradio.my/nightdx-unattended-shortwave-medium-wave-listening-airspy/

#unattendedovernight #mediumwave #wavelistening #overnightshortwave #nightdx #quicksummary #marcozennaro #qsl #every #python

#rss #feeds #linksharing #qwibbit

Telescope: Turn an Android Phone into a Pro Webcam for Linux and ...

2026-10-30T15:04:26Z

Telescope: Turn an Android Phone into a Pro Webcam for Linux and Windows

TL;DR: Quick Summary
Telescope by LunarKittyy is a free and open source AGPL-3.0 application that turns an Android phone into a pro webcam for a Linux or Windows computer. Every real lens works (ultra-wide, main, telephoto), ISO, shutter, white balance, focus and torch are adjustable live from the desktop, and the phone mic can serve as the computer mic. Video travels phone to computer only, encrypted with TLS, over USB or Wi-Fi, with no account, no cloud and no tracking. Version 4.4.2 adds ten new interface languages, per-phone pairing buttons, webcam and capture-card inputs, and a fix that blocks a MediaPipe usage report. For amateur radio operators this solves the classic shack video problem: a beat-up laptop webcam becomes a sharp phone camera for club talks, remote exam proctoring, build livestreams, net check-ins with a bench view, and Field Day documentation, while desk mode flattens logbooks and schematics for the call and the On Air sign can light a shack lamp through a webhook.
What Is Telescope?
Telescope is a free open source phone-as-webcam system that streams video and audio from an Android phone to a Linux or Windows PC, where it appears as an ordinary virtual webcam and microphone any app can use.
Install the desktop app on the computer and the phone app on Android, pair once with a QR code or USB cable, then pick Telescope (Windows) or Phone Camera (Linux) in Discord, Zoom, Teams, OBS Studio or a browser. The desktop handles lens choice, exposure, framing and effects as plugins, most of them optional and off until wanted. Up to four phones, tablets, laptops or USB webcams can stream at once, each on its own virtual camera, or combined into one picture for apps that take a single camera. Anything else with a browser (iPhone, tablet, old laptop) can stream too with nothing to install, through the browser-camera page.
The project lives at https://github.com/LunarKittyy/Telescope with the product site and comparison at https://telescope.webcam. It holds about 48 stars and 1 fork, 1,445 commits on master, and ships a Windows installer plus portable zip, a Linux tarball with start.sh, and an Android APK, all from the GitHub releases page.
Telescope desktop app streaming from an Android phone, with the phone app beside it. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Project Snapshot
Item
Detail
Project
Telescope by LunarKittyy
License
GNU AGPL-3.0, free and open source
Latest release
4.4.2 (28 Oct 2026: ten new languages, pairing race fix, MediaPipe telemetry block; 4.4.0 added webcam inputs; 4.3.0 added thirteen plugins and Windows 11 camera)
Desktop platforms
Windows 10 and 11 (per-user installer, no admin; portable zip), Linux (tarball with start.sh, Python 3.12 plus)
Phone platforms
Android 8 plus native app; any other phone, tablet or laptop via browser camera with nothing to install
Codebase
Python desktop with Qt interface and plugin host, Kotlin Android app, WebRTC browser path, UnityCapture driver on Windows, v4l2loopback on Linux
Network
USB via adb or Wi-Fi via TLS; experimental UDP option; browser camera over WebRTC on LAN port 8767; On Air page on port 8768
Privacy model
No account, no cloud relay, no tracking in the apps; video stays phone to computer; update checks hit GitHub only; MediaPipe vision telemetry blocked since 4.4.2
Links
GitHub repo, product site, full feature docs, comparison, releases
Why Phone Cameras Beat Laptop Webcams
A built-in laptop camera is usually a fixed-focus 720p module with a tiny sensor, no low-light headroom and a wide-angle lens that flatters nobody. A mid-range Android phone from the last five years carries a larger main sensor with autofocus, a real ultra-wide and often a telephoto, optical or electronic stabilization, a torch, and a DSP that handles noise far better.
The established phone-as-webcam tools each keep something back. DroidCam watermarks HD unless paid and gates manual controls and ad removal behind Pro. Iriun watermarks the free tier and gates manual ISO, exposure and white balance. Camo caps free output at 720p and gates 4K, 60 fps and Smart Zoom. All three are closed source with per-vendor accounts, cloud hooks or ad SDKs in the loop.
Telescope takes the opposite stance: the whole app is free, every lens and manual control is included, output runs up to 4K, there is no watermark and no ads, and the source is public under AGPL-3.0. The trade is narrower platform coverage (Android plus Windows/Linux only, no native iPhone or Mac app) and a younger device-testing base than tools that have been around for years.
How It Fits Together
Three pieces cooperate, and each has a clear job.
1. The Android app
The phone app owns the camera hardware. It enumerates the real wide, main and telephoto sensors, exposes ISO, shutter, brightness compensation, white balance, focus, stabilization, noise reduction and torch per lens, and encodes the stream as H.264 (Light format) or MJPEG (Heavy format). It keeps screen and Wi-Fi awake while open to avoid lag spikes, drops to 720p when very hot on Android 10 plus, shows as a Live Update with Stop button on Android 16 plus, and can run with the screen blacked out to save OLED battery. Pairing tokens live in a config folder readable only by the user on Linux, and the desktop pins the phone certificate at pairing so a different certificate is refused.
2. The desktop app
The desktop app owns the virtual camera, the mic path, the plugin pipeline and pairing. On Windows it installs a UnityCapture-based driver (with a dedicated Windows 11 camera since 4.3 so every app including the Windows Camera app can see it). On Linux it drives v4l2loopback through PulseAudio or PipeWire, exposing Telescope Microphone with no extra packages on most desktops. The interface is card based (Camera, Transforms, Stream output, Microphone, Connection plus one card per enabled plugin), cards fold, everything is keyboard navigable, and almost any control can be bound to a key. Presets save camera, output, transform and active plugin settings per phone, lens included, each with an optional shortcut.
3. The link between them
USB when the phone is plugged in and answering over adb, Wi-Fi otherwise, with a per-phone override to force either. A new router IP breaks nothing because the phone announces itself on the network. Streams reconnect across transports (pull the cable mid-call and it continues over Wi-Fi), apps see a configurable wait screen after 10 seconds of loss instead of a frozen frame, and a banner names the phone as unreachable after 30 seconds. Light format defaults to about 8 Mbps at 1080p30 with Auto or Dynamic bitrate that backs off before lag starts and even drops frame rate to 15 before giving up, then climbs back. Heavy format trades bandwidth for sharper fast motion and suits USB or strong Wi-Fi. An experimental UDP switch trades reliability signals for lower latency on crowded airwaves but can stutter on many networks, which is why 4.4 moved it from the Connection card into Settings where it is harder to flip by accident.
The guided checklist sets up the virtual camera, installs the phone app over QR and pairs the two. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Up and Running in About Two Minutes
Download the desktop build from the releases page: TelescopeSetup.exe on Windows (per-user, no admin needed) or Telescope-linux.tar.gz on Linux (extract and run ./start.sh, which installs the pinned Python dependencies and launches).
Follow the checklist that opens on first run. It sets up the virtual camera (one password prompt on Linux for the driver, one Update click on Windows if migrating to the Windows 11 camera), puts the phone app on Android over a QR code, and pairs the two. With the phone on USB, 4.4.1 and later show a Pair button per phone instead of pairing immediately, so a cable-left-plugged phone can’t win the race against the one being scanned.
Pick the camera in the meeting or streaming app: Telescope on Windows (Telescope 2 to 4 for extra streams), Phone Camera on Linux. Set the mic to Telescope Microphone (Linux) or CABLE Output after the one-time VB-Cable install (Windows).
What Linux needs: a v4l2loopback-capable kernel path handled by start.sh, PulseAudio or PipeWire for the mic, and optionally VA-API setup for GPU decode of Light streams on AMD and Intel (NVIDIA works out of the box, Windows uses any GPU). No Android phone at all still works: pick Browser camera in the phone picker, scan the code from an iPhone, tablet or old laptop, accept the self-signed certificate warning once, tap Start, and keep the page open with the screen on.
Camera Control: Every Lens, Set by Hand
The Camera card is the reason to pick Telescope over a generic reflector.
Lens picker lists the real wide, main and telephoto sensors instead of faking it with crop zoom. Auto mode switches lenses as the framing demands.
ISO and shutter by slider or typed value, ranged per lens, plus brightness compensation (usually plus or minus 8 EV).
White balance in Kelvin from 2000 to 10000 with green-magenta tint, handy under mixed shack fluorescents and warm bench lamps.
Focus in three modes: Auto, tap-a-point (click the preview to focus and meter there), or Manual slider. Greyed-out controls mean the current lens can’t do them; hovering the control header shows per-lens capability.
Helpers: stabilization, noise reduction, sharpening, black level lock and torch where the lens offers them, plus a Low light switch that lets auto exposure run the sensor slower (often down to 10 fps at the dark end) so each frame gathers more light, but only when needed.
Zoom 1 to 10x with pan on the Transforms card (up to 30x with Max zoom in Settings), flip and 90-degree rotation, independent width and height stretch from 50 to 200 percent. Zoom crops from the full sensor first and hops to the telephoto once the framing fits inside it, which is why zooming in gets sharper instead of blurrier. Dots on the slider mark each lens handover.
ISO, shutter, focus and white balance set by hand from the desktop, Pro mode style. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Stream Pipeline: Resolution, FPS, Format and Bandwidth
Resolution list comes from what the current lens and encoder really support, up to 4K. One FPS setting (15 to 60) drives phone and virtual camera together, with unsupported rates greyed out.
Canvas size in Settings decouples the virtual camera from the phone: 720p, 1080p or 4K in landscape or portrait, 4:3, or custom even-numbered sizes. On Linux a canvas change reloads the driver (close OBS first).
Light (H.264) suits most links at roughly 8 Mbps for 1080p30. Bitrate modes are Auto, fixed 1 to 100 Mbps, or Dynamic which rides the link capacity and trims frame rate before quality collapses. Hardware acceleration decodes on the GPU, most useful at 4K or on laptops.
Heavy (MJPEG) suits sharp fast motion on USB or strong Wi-Fi, with a JPEG quality slider. Phones that can’t do Light fall back to Heavy automatically.
Footer telemetry shows FPS and throughput live, with a Can’t keep up note that distinguishes weak signal or 2.4 GHz congestion from a full dropout, so the fix (move closer, switch to 5 GHz, plug in USB) is obvious.
Microphone Without a Second Device
The phone mic becomes a computer mic while streaming.
On Linux, Telescope Microphone appears through PulseAudio or PipeWire with nothing extra to install on most desktops.
On Windows, the card offers a one-click VB-Cable fetch from VB-Audio (admin prompt plus restart), after which CABLE Output is selected as the mic in other apps.
Gain spans minus 24 to plus 12 dB with a limiter (yellow) and clip (red) meter, optional Auto level that rides quiet and loud speech without lifting room noise in pauses, and optional RNNoise-based noise reduction running on the computer (about 20 ms extra delay, best with the call app’s own suppression switched off to avoid thin double-processed audio).
Mute lives on the card and in the tray, resets on restart, and can be bound to a push-to-talk key. Camera-off with mic-on is supported (apps see the wait screen), including a start-mic-only mode.
The Plugin System: Switch On Only What You Want
Most of the desktop is plugins now, a design that landed across 3.5 through 4.3. The Plugins tab in Settings groups them, each with its own search, per-plugin restart, and safe handling (a changed or crashing plugin is left out with a banner naming it). New plugins install from a zip by dialog or drag and drop, start switched off, and carry an explicit trust warning since a plugin can run arbitrary code. Settings and presets survive toggling.
Auto framing tracks the presenter with zoom and pan and widens for two people. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Auto framing and Expose for my face
Auto framing finds people with the shared MediaPipe vision download (about 47 MB on Linux, 28 MB on Windows, fetched once from GitHub releases) and drives zoom and pan to hold the framing, from head-only to full body, framing two people together and zooming out after two seconds with nobody in view. Speed (Sloth to Zippy, Easy default), room-to-move deadband (Fidget to Roam, Off follows everything), and an optional widen-for-something-in-hand mode tune the behavior. A skeleton overlay shows what tracking reacts to, preview only, never in the output. Power-saving effects in Settings slows vision plugins while moving to save CPU and battery at the cost of slightly later following.
Expose for my face moves the phone metering point to the face, which rescues backlit shack windows, waits during manual focus, and parks where the face was when leaving frame instead of chasing the empty room.
Background and Chroma key
Background (vision-based) blurs the room or replaces it with a PNG, JPEG, WebP or BMP, with edge softness control, running on the computer so it works with any phone and browser camera. Chroma key (no download) swaps a green, blue or picked-color screen for a color or picture with tolerance, edge and spill controls, giving sharper hair edges under even light at far lower CPU than segmentation.
Chroma key swaps a green or blue screen for a picture with no model download. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Desk and document mode
Point the phone down at a page, drag four corners onto the sheet on the preview, and Telescope rectifies it flat and head-on for the call. Shape modes (Auto perspective, A4, Letter, Square, Fill), Contrast mode for uneven light, and Black and white mode for print and pen turn a shaky overhead shot into a readable document. Tap-to-focus still lands on the clicked word, each phone remembers its corners, and the card warns if zoom or framing pushes the sheet out of frame.
Desk mode rectifies paper, sketches and whiteboards to face the call head on. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Several cameras in one picture
With two to four streams running, the One picture plugin composites them side by side, in a grid, big-and-small, or in a free layout with OBS-style drag, snap, Shift-stretch, Alt-crop, per-stream zoom window, borders and rounded corners over a color or picture. The composite goes out on the main Telescope camera while each stream still feeds its own virtual camera for OBS scenes. Layouts save by name, switch by key, and ride along in presets. Streams that drop leave the picture after 3 seconds and rejoin on reconnect without freezing the rest.
Several cameras in one picture for apps that accept only one camera. Image credit: LunarKittyy, Telescope repository, AGPL-3.0.
Fun pack, Colour grading and Cinema tools
Fun pack (under Extras) bundles Clone party up to eight copies, Invisibility cloak against a captured empty room, Light painting trails, Slit-scan motion shear, Stick figure tracking, retro Game Boy, 1998 webcam, VHS, CRT and ASCII looks, and Datamosh keyframe-drop smear on Light streams. Colour grading adds Lift, Gamma, Gain and Offset wheels plus temperature, contrast with pivot, saturation, shadows and highlights, RGB and Hue curves, .cube LUT import with strength, grade export, per-phone grades and a before-after split. Cinema tools adds focus peaking, zebras with adjustable threshold, false color, thirds and golden-ratio guides, histogram, waveform, RGB parade, vectorscope and a sensor-dr

https://hamradio.my/telescope-phone-as-webcam-android-linux-windows/

#androidphone #prowebcam #windows #linux #telescope #desktop #camera #quicksummary #turn #usb

#rss #feeds #linksharing #qwibbit

Ollee Watch: The Drop-In Bluetooth Brain That Turns a Vintage ...

2026-10-29T06:04:42Z

Ollee Watch: The Drop-In Bluetooth Brain That Turns a Vintage Casio Into a Smart Watch

TL;DR: Ollee Watch (https://www.olleewatch.com/) is a $55 drop-in replacement circuit board for vintage Casio watches on the 593 module family, including the F-91W, F-84, A158, and A159. No soldering is needed. You swap the PCB, pair the free phone app, and gain Bluetooth time sync, world time, sunrise and sunset, 5-day weather, step counter, RGB backlight, NFC tag, and wireless firmware updates while keeping the iconic look. A pre-assembled watch costs $99 when in stock. Battery life runs 10 months daily or 3 plus years in power saver mode.
What Is Ollee Watch?
Ollee Watch is a drop-in Bluetooth replacement mainboard that converts a classic Casio F-91W, A158, or other module 593 watch into a connected smart watch without changing its iconic look. Open the case, lift out the original Casio PCB, transfer the battery contact, drop in the Ollee board, close it up, and pair with the free Android or iOS app. The familiar 7-segment LCD stays, but behind it now sits an ARM Cortex-M4/M0+ processor with Bluetooth Low Energy, an accelerometer, a loud buzzer, an RGB backlight, and an optional programmable NTAG213 NFC tag. There are two flavors: Ollee Watch One with RGB LED backlight and NFC support, and Ollee Watch EL with an electroluminescent panel and 4 brightness settings.
Why the Humble F-91W Deserves Superpowers
The Casio F-91W has been in continuous production since 1989, sells for around $20, survives about 3 million units a year, and runs up to 7 years on one coin cell. It’s the watch of presidents and pop culture, a field-work legend, and the default beater for hams, hikers, and makers. Its weakness was never the case or the display. It was the frozen-in-1989 feature set: no sync, plus or minus 30 seconds drift a week, one alarm, green-only backlight, and zero awareness of where on Earth you are.
Ollee Watch keeps everything people love (weight, size, pushers, water resistance of the unmodified case, the honest LCD) and replaces only the brain. The result still looks, wears, and sips power like a Casio, but behaves like a modern connected instrument.
One vs EL: Which Board to Buy
Feature
Ollee Watch One
Ollee Watch EL
Backlight
RGB LED, color and duration configurable, 3-mode flashlight
Electroluminescent panel, 4 brightness settings
NFC tag (NTAG213, programmable)
Optional, resin F-91W only, not A158W
Not available
Processor and radio
ARM Cortex-M4/M0+, Bluetooth LE
ARM Cortex-M4/M0+, Bluetooth LE
Accelerometer and gestures
Yes, step counting plus tilt-to-wake auto light
Yes
Buzzer
Loud alarm buzzer
Loud alarm buzzer
Battery
CR2016 coin cell
CR2016 coin cell
Battery life
10 months daily use, 3 plus years power saver
10 months daily use, 3 plus years power saver
Best for
Tinkerers, NFC experiments, color fun
Readers who want soft even EL glow
Both boards work through the same phone app with free Android and iOS builds, wireless firmware upgrades, and sync with Apple Health and Google Health for steps. Official compatibility covers the F-91W, A158W, F-105W, and A168, and the community has fitted the F-84W, A159W, A171W, and W-59 since they share similar module 593 layouts.
Features That Matter on the Wrist
Bluetooth time sync: zero drift per week once paired. The watch follows your phone instead of wandering 30 seconds weekly.
World time: multiple zones a few button presses away, set from watch or app. Handy for scheds, nets, and DX windows.
Sun, moon, and tide: sunrise and sunset for your location, moon phase, tide times. Set location once in the app.
5-day weather: daily high, low, feels-like, conditions, precipitation, humidity, and UV index on the small screen.
Step counter: accelerometer-based counting with daily goal and alert, stored 30 days in the app.
Alarms, timers, stopwatch: configurable day alarms with chime, snooze, and hourly beep, up to 10 programmable timer presets with interval and fast modes, and a 1/100-second stopwatch with logging up to 24 hours.
Tally counter, temperature display, databank: everyday utilities that suit field logging and contest counting.
Flashlight and games: 3-mode backlight flashlight plus character-based BlackJack and Ping tech-demo games.
Customization: watch faces, app order in the mode cycle, backlight color and afterglow, battery saver, tilt-to-wake auto light.
Heart rate helper: no optical sensor. This is a pulsometer face. Count 20 beats on your wrist, stop the timer, and the watch computes BPM from the elapsed time.
The app holds steps and timed events for up to 30 days, pushes alarms and time zone settings to the watch, and handles firmware updates over the air.
Installation: No Soldering, Just Patience
StrapHabit reviewer Nick Lehner documented the full swap into an A158, and Ollee publishes step-by-step instructions plus a YouTube video at https://www.olleewatch.com/get-started-one. The honest summary: if you can change an A158 battery, you can do this mod.
Gather a small Philips screwdriver, a flathead screwdriver, and tweezers. No tools ship in the box, which holds the module, NFC chip, and a sticker.
Remove the 4 caseback screws and lift the back. Work on a clean tray so tiny parts can’t roll away.
Separate the LCD from the original circuit board, exactly as during a battery change.
Transfer the battery contact to the Ollee board. This is the fiddliest step. A small screwdriver and tweezers make it easy.
Drop in the Ollee board, skip the original Casio PCB entirely, and reassemble in reverse. Skip the NFC chip if you have no use for it, since it complicates the stack.
Install the phone app, pair over Bluetooth, set location and preferences, and sync time.
Total hands-on time is well under an hour on a first attempt. Take a phone photo of the original stack before lifting anything so the contact orientation is on record.
Battery Life: The Number That Sells It
Smartwatches that demand a charger every night trained everyone to expect power anxiety. Ollee Watch instead runs a CR2016 coin cell for about 10 months of daily connected use with default settings, stretching past 3 years in power saver mode. That is shorter than the stock 7-year Casio cell, but it is in a different universe from rechargeable smartwatches. For EDC, field days, and travel, a watch that you never plug in wins.
Ollee Watch vs the Alternatives
Option
Price
Display
Sync
Standout
Catch
Stock Casio F-91W or A158
About 20 to 30 dollars
Original LCD
None, drifts weekly
7-year battery, iconic, cheap
Frozen 1989 feature set
Casio F-91W plus Ollee Watch DIY kit
About 85 dollars total
Original LCD
Bluetooth phone sync
RGB or EL light, weather, sun times, steps, OTA updates
Install it yourself, 10-month battery
Pre-assembled Ollee Watch
$99
Original LCD
Bluetooth phone sync
Zero surgery
Frequently out of stock
Sensor Watch and Sensor Watch Pro
From $35 board
Original or upgraded display
None by design
Fully programmable faces, open firmware
No phone sync, you write the code
Casio ABL1000WE
About $95
Similar steel look
Bluetooth
Factory warranty, no install
Larger 41.6 mm case, fewer smart extras
StrapHabit concluded the kit costs nearly double the donor A158 and lands near watches with native step counting, but none of those preserve the small classic case plus the mod community fun. If you want factory polish, buy the ABL1000WE. If you want the vintage silhouette with modern brains and you enjoy the install, Ollee wins.
A Ham Radio Field Watch?
This is hamradio.my, so here is the field angle. Sunrise and sunset on the wrist helps plan greyline DX without pulling out a phone. World time tracks UTC plus the DXpedition zone. The interval timer paces antenna tests, satellite pass drills, and contest exchanges. The tally counter logs contacts at a demo table. The loud buzzer actually wakes you for a 4 AM sched. And 10-month battery means the watch is alive when the Field Day box opens, unlike the dead smartwatch in the same tote. Pair it with a NATO or FKM rubber strap for sweat and rain, and it becomes a proper SOTA and POTA companion.
Honest Limitations Before You Buy
Supply is limited and the $99 pre-assembled version sells out. The DIY kit can take weeks to arrive.
NFC works on resin F-91W builds only, not on A158W metal-case models.
The heart rate face is manual pulse math, not a sensor. Fun, not medical.
No third-party app store. Games are character tech demos, not Apple Watch apps.
Language purists should note the 7-segment LCD can’t show rich text, and some commenters wish for a 16-segment upgrade.
Battery swaps mean reopening the case, so keep spare CR2016 cells and respect the gasket if you care about water resistance.
Frequently Asked Questions
What is Ollee Watch in one sentence?
Ollee Watch is a $55 drop-in Bluetooth mainboard that adds phone sync and smart features to classic Casio watches.
Which Casio models fit the Ollee Watch?
F-91W, A158W, F-105W, and A168 officially, plus F-84W, A159W, A171W, and W-59 by community testing.
Do I need to solder anything?
No. The swap uses screwdrivers and tweezers only, with no soldering required at any step.
How long does the battery last?
About 10 months with daily connected use, or more than 3 years in power saver mode on CR2016.
Is a pre-assembled Ollee Watch available?
Yes, a $99 pre-installed F-91W is sold on the store page, but it is frequently out of stock.
Sources and Further Reading
Ollee Watch official store: https://www.olleewatch.com/
Ollee Watch get started and installation guide: https://www.olleewatch.com/get-started-one
Liliputing coverage: https://liliputing.com/ollee-watch-brings-smartwatch-features-to-the-casio-f-91w-with-a-mainboard-replacement/
StrapHabit review and installation in Casio A158: https://straphabit.com/blogs/straphabit-reviews/ollee-watch-review-and-installation-in-casio-a158
Ollee Watch companion app on Google Play: https://play.google.com/store/apps/
TechEnclave owner discussion: https://techenclave.com/t/ollee-watch-vintage-casio-f91-a158-mod/401496
Sensor Watch open programmable board: https://www.tindie.com/products/joeycastillo/sensor-watch/
Casio F-91W product page: https://www.casio.com/us/watches/casio/product.F-91W-1/
73
The post Ollee Watch: The Drop-In Bluetooth Brain That Turns a Vintage Casio Into a Smart Watch appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/ollee-watch-vintage-casio-bluetooth-smart-mod/

#vintagecasio #nfc #casiopcb #backlight #battery #arm #buzzer #loud #featureollee #board

#rss #feeds #linksharing #qwibbit

QRZero: A Fast Local First Amateur Radio Logger for Windows, ...

2026-10-20T08:44:39Z

QRZero: A Fast Local First Amateur Radio Logger for Windows, macOS and Linux with DX Chasing Built In

TL;DR: QRZero by nreed97 is a free open source desktop amateur radio logger for Windows, macOS (Apple Silicon) and Linux that keeps your whole log in a single SQLite file, stays responsive past 200,000 QSOs, and integrates the full shack: WSJT-X, JTDX and N1MM over UDP, Hamlib and serial rig control, rotators, DX cluster with Needed now alerts, DXCC through CW club awards, LoTW, QRZ, Club Log and eQSL uploads, and paper QSL label printing. Built with DX chasing and CW in mind, local first with no cloud account, and honest about its demo screenshots using the placeholder call N0CALL.
Every ham eventually hits the same wall with logging software. The log grows past 50,000 QSOs and the grid starts stuttering. The DX cluster lives in one program, awards tracking in a spreadsheet, LoTW uploads in another window, and the CW club numbers in a paper notebook. Contest season means exporting and importing ADIF files and hoping no field gets mangled in transit.
QRZero was built to collapse that whole pile into one window that never makes you wait mid QSO. The screenshot above shows it mid session: a New QSO entry for OH2BH on 20m CW, a Station pane that instantly says Finland is already worked on this band and mode with the full DXCC slot grid, a grey line map with beam headings, a Worked before history, and a log grid holding 1,434 contacts without breaking a sweat.
What Is QRZero?
QRZero is a fast local first amateur radio logging application for Windows, macOS and Linux that stores the log in one SQLite file and combines QSO entry, rig control, digital mode integration, DX cluster chasing, awards tracking and QSL management in a single desktop window.
The stack is modern: a Rust workspace for the log database, ADIF handling, rig control and cluster client, plus a React and TypeScript UI in a Tauri shell, with the same screens also servable to a web browser from a small local server. Passwords live in the OS credential store (Windows Credential Manager, macOS Keychain, Linux desktop keyring), never in the log file, and the app only touches the internet for services you explicitly switch on.
Four design goals define it:
Speed. The grid renders only visible rows, searches are indexed, and the interactive queries are tested against a 200,000 QSO log. The stated goal is a logger that never makes you wait mid QSO.
Every mode is equal. CW, phone and digital get identical treatment in entry defaults, awards, band map and spot colours. No second class modes.
Nothing is lost. Every ADIF field of every QSO is stored, including proprietary APP fields other programs add, and a Full export returns them all. One file, with daily automatic backups.
Local first. No account, no cloud, no wizard to fight. The classic dense desktop logger layout with dockable panes you can move, tab, resize and pop out to a second screen.
Logging QSOs the Fast Way
Type the call, press Tab or Space, and QRZero looks it up and fills name, QTH, state, county, grid, country and DXCC. Press Enter to log. Start time is captured when you begin typing, the form clears, and the cursor returns to the call box. Lookup filled boxes render in a different colour from typed text, and typed text always wins.
Lookups use QRZ.com XML with a subscription (cached 30 days) and fall back to your own earlier QSOs and the country file, with Fill from QRZ and View on QRZ.com buttons beside Clear.
Frequency, band and mode follow each other. Type a frequency and the band sets itself. FT4 stores the ADIF correct way (MFSK with FT4 submode) while displaying as FT4.
Entry fields are fully yours. Pick a starting layout (general DX, CW, CW clubs, parks and summits, satellites, casual contesting), then add, remove, rename, resize or default any ADIF field, or invent your own. Tick keep to carry a value across QSOs. CW, Phone and Digital can each own a layout, so CW shows club number boxes while SSB shows something else.
CW club numbers get a dedicated layout with SKCC, NAQCC, FISTS and CWops boxes that prefill from your last QSO with that call and feed the club awards engine.
Split aware. With a split radio the form shows TX and RX boxes, both are logged, and spotting uses the frequency you listened on. Cluster comments like UP 5 or QSX 14.205 auto configure split when clicked.
Rig, antenna and amplifier per location. Antennas carry their band coverage so the Ant box can auto select. Each QSO records MY_RIG, MY_ANTENNA and power.
Spotting from the entry panel shows the exact cluster line before anything is sent, and only QSOs from the last 10 minutes qualify, so stale contacts never get spotted by accident.
The Worked before pane lists every earlier QSO with the station including portable variants, with duplicate markers and band, mode and unconfirmed filters. The Station pane shows what the contact is worth for every award plus a band by mode DXCC slot grid (W worked, C confirmed, blank needed). Station notes (“QSL via bureau only”, “sked on 40m”) follow the callsign and are searchable. The Map draws short and long paths with beam headings and distances, flat or azimuthal projections, live grey line shading with sunrise and sunset at both ends, and Turn SP and LP buttons when a rotator is connected.
Radios and Rotators
Each radio is equipment at a location with its own control method:
Rig control
Covers
Hamlib rigctld
Almost any radio Hamlib supports
TCI
ExpertSDR, SunSDR, Thetis, Flex via bridge, every slice as its own radio
Kenwood, Elecraft, Flex CAT
Serial or SmartSDR CAT ports
Yaesu CAT
FT-991A, FTDX10, FTDX101, FT-710 and other recent Yaesu rigs
Icom CI-V
IC-7300, IC-7610, IC-705 and others over USB or CI-V interface
Follow mode ties frequency, band and mode to your tuning both directions, and clicking decodes, spots or DXpedition rows tunes the radio. Multi radio operation covers SO2R and Flex slices with Alt+1 and Alt+2 switching and a red readout on transmit. Rotators connect over Hamlib rotctld, Yaesu GS-232 serial or TCP, or PstRotatorAz UDP, with N1MM bearings accepted as input. The Rotator pane is a clickable compass dial with presets, short and long path buttons for the entry callsign, and an optional azimuthal map backdrop.
WSJT-X, JTDX and N1MM Integration
QRZero listens to any number of WSJT-X and JTDX instances (separate ports or shared multicast) plus N1MM Logger+ over UDP. Contacts logged in those programs land in your log at the chosen location, with QRZ data backfilled and nothing they sent overwritten. A Pass on relay feeds GridTracker or JTAlert from the same stream.
The FTx monitor merges every running decoder into one list flagged against your log: New DXCC, New band, New mode, New grid (6m and up), New call, New on band, Worked, plus callers calling you and watch list hits. Click fills the QSO panel without retuning, double click triggers WSJT-X to call. Per instance control strips show TX state with Halt TX and Call CQ, a call boxes view mirrors the JTAlert style layout coloured by top alert, and filters hide worked stations or set signal floors. N1MM radio info can drive the QSO panel and N1MM can steer your rotator through QRZero.
Chasing DX: Cluster, Needed Now, Band Map and More
The DX cluster client connects through your node list with failover, flags spots against your log exactly like FTx decodes, colours by mode (blue CW, green digital, orange phone), replaces duplicates, and filters by band, mode group, age, Needed only and Hide worked, with origin filters for impossible paths. Clicking a spot fills the form and tunes the radio, and a console speaks raw node commands.
The Needed now pane is the DX chaser’s home screen: only spots that would be new for your log, ranked new entity first, then new band, then new mode, each group ordered by the entity’s rank on Club Log’s public Most Wanted list (fetched daily, no login). Sound alerts (double beep for new entity, single for band or mode) and corner popups that tune on click are independently switchable, with per call quiet periods and sensible scope settings.
Around it sit a full toolkit: a zoomable band map with cluster spots, FT8 decodes and band plan strips. A watch list for calls, prefixes or entities with top bar notices; an NG3K DXpedition calendar marking what is new for you with click to tune; a WA7BNM contest calendar with mode filters and rules links; and a propagation pane with flux, sunspots, A and K, solar wind and N0NBH band estimates plus a per call 160m to 10m 24 hour band forecast computed locally from solar data and path geometry.
Awards from DXCC to CW Clubs
The Awards pane tracks DXCC (with Challenge, 5BDXCC and optional LoTW verified credit), WAS, WAZ, WPX, WAC, ITU zones, VUCC grids on 6m and up, IOTA and US counties. Rows show Mixed, CW, Phone and Digital plus every band 160m to 70cm with green C confirmed and amber W worked cells, configurable confirmation sources, per callsign limits, and click through to the underlying QSOs. Unwanted awards switch off cleanly without touching the log.
CW operators get rare first class treatment: SKCC (Centurion, Tribune, Senator with dates and single band endorsements), CWops (ACA, ACMA, CMA, member DXCC and states, plus CWT medals with per hour breakdowns and manual points for other logs), NAQCC Friendship Club levels, and FISTS Century tiers through Diamond plus WAS, all counted from logged club numbers by member number. The honest caveat is stated in the docs: QRZero holds no member lists, so validate against club rules before applying.
QSL and Online Services in One Window
One QSL window covers the paper and digital worlds. The Queue tab handles cards: queue to send, mark received with reply queuing, contact info per station, Avery and Brother label layouts (including direct to printer Brother QL output and PNG saves), bureau grouped sorting, and sent via bureau or direct marking. Online services get one page each with its own credentials and auto upload timers: LoTW signs through your installed TQSL with background confirmation downloads that backfill states, zones, grids and counties; QRZ Logbook, Club Log and eQSL auto upload new and edited QSOs minutes after the last change with date range manual uploads and on demand confirmation downloads. A New toward your awards list after each download shows exactly what the fresh confirmations earned. Importing from another logger? Download confirmations first so nothing already confirmed gets re sent.
Your Data, Your Window, Your Machine
ADIF import keeps every field, links QSOs to a location, skips duplicates safely and absorbs unknown station callsigns. Export writes selected rows, the current grid view, or full filter matches in Standard (for other loggers and services) or Full (for backups and QRZero to QRZero moves) fidelity. Multi log and multi callsign support shares locations and equipment, archived callsigns retire gracefully while keeping history, and the Statistics window summarizes totals, yearly DXCC progress, per service confirmation rates and top entities in about half a second on 200,000 QSOs.
The window itself is fully yours: dockable tabbed pop out panes with Always on top and named layouts, light and dark themes, 11 to 18 px text, remappable keys with printable cheat sheets, and a search box over all settings. The same UI runs in a browser from a local server binary, UDP connections feed anything else in the shack with templated N1MM, ADIF, JSON and PstRotatorAz messages, and startup programs launch your helpers automatically. Daily backups keep ten copies with verified restores and a pre restore safety copy.
QRZero vs Other Logging Options
Feature
QRZero
Cqrlog
KLog
N1MM+
Cloudlog
HAMRS
Price and license
Free, open source
Free, open source
Free, open source
Free
Free, open source
Free tier plus paid
Platforms
Windows, macOS ARM, Linux
Linux mainly
Win, macOS, Linux
Windows only
Any browser, self hosted
Win, macOS, Linux, mobile
Log storage
Single local SQLite file
MySQL or MariaDB local
SQLite
Local files
MySQL, self hosted
Local plus cloud sync
200k QSO performance
Tested design goal
Can slow without tuning
Good
Excellent
Depends on server
Good
Digital integration
WSJT-X, JTDX, N1MM UDP
WSJT-X via ADIF
WSJT-X direct
Native contest engine
WSJT-X API
Manual plus WSJT-X import
DX cluster plus alerts
Yes, Needed now with sound
Basic cluster
Basic cluster
Yes, contest focused
Via addons
No
Awards depth
DXCC to CW clubs with medals
DXCC basics
DXCC, WAS, WAZ
Contest scoring
DXCC via addons
POTA, SOTA focused
QSL services
LoTW, QRZ, Club Log, eQSL plus paper
LoTW, eQSL
LoTW, eQSL, Club Log
Limited
LoTW, eQSL via cron
Limited
Internet required
No, local first
No
No
No
Yes, server based
Partially
QRZero’s niche is clear: the offline desktop DX chaser and CW operator who wants one fast program instead of a chain of tools, without surrendering data to a server.
Honest Limitations
Windows is the main platform and gets the most testing. MacOS and Linux builds exist but trail it.
macOS is Apple Silicon only and unsigned. No Intel build, and first launch needs Control click Open or an explicit Privacy and Security approval since there’s no Apple Developer signature.
No club member lists. CW club awards count from your logged numbers, so confirm eligibility against club records before applying.
Contest calendar only. The WA7BNM pane lists contests with rules links, but live contest logging stays in N1MM with QSOs flowing to QRZero over UDP.
Browser server needs a source build. Release downloads carry the desktop app only. The local web server binary is built from source with Rust and Node 22.
Forecast is a guide. The band forecast is computed locally from flux, K index, sun geometry and date. It knows nothing about sporadic E and replaces no prediction program.
Frequently Asked Questions
Which platforms does QRZero run on?
Windows with installer, MSI and portable builds, Apple Silicon macOS 11 and up via DMG, and Linux via deb, rpm or AppImage. Windows receives the most testing.
Can it handle a large existing log?
Yes. Import ADIF with duplicate skipping, every field preserved, and a grid and queries tested against 200,000 QSOs with daily automatic backups keeping ten copies.
Does it work with WSJT-X and my radio?
Yes. WSJT-X, JTDX and N1MM connect over UDP with an FTx monitor flagging decodes against your log, and rigs connect via Hamlib, TCI, serial CAT or CI-V with SO2R and rotator support.
Which awards and QSL services are covered?
DXCC with Challenge, 5BDXCC and verified credit, WAS, WAZ, WPX, WAC, VUCC, IOTA, US counties and SKCC, CWops, NAQCC and FISTS awards, plus LoTW, QRZ, Club Log, eQSL and paper card label printing.
Is my log data safe and private?
Your log is one local SQLite file, passwords stay in the OS keyring, networking only runs for services you enable, and the daily GitHub update check downloads nothing and can be disabled.
Sources and Further Reading
QRZero GitHub repository
QRZero releases page
QRZero user guide
ADIF amateur data interchange specification
ARRL Logbook of the World
QRZ Logbook
Club Log
eQSL.cc
WSJT-X weak signal software
N1MM Logger+
Hamlib rig control
NG3K DXpedition calendar
WA7BNM contest calendar
SKCC Straight Key Century Club
CWops
73 de 9M2PJU
The post QRZero: A Fast Local First Amateur Radio Logger for Windows, macOS and Linux with DX Chasing Built In appeared on HamRadio.my - Amateur Radio, Tech Insights and Reviews by 9M2PJU.

https://hamradio.my/qrzero-fast-amateur-radio-logger-windows-macos-linux/

#amateurradio #radiologger #localfirst #fastlocal #linux #qso #windows #log #built #qrzero

#rss #feeds #linksharing #qwibbit