IZ6RND/B

The idea of building a beacon never began with a formal design or a technical ambition. It began with late-night conversations with my friend Domenico Berrè, IK6IHH - Mimmo - a trusted companion in projects that usually start with a question and end somewhere between experimentation and adventure. We often discussed APRS, antennas, unusual circuits, and half-serious ideas that always deserved at least one attempt. Somewhere during these talks, the concept of a "Radiofaro" emerged: not just a transmitter, but a presence - something that speaks slowly, precisely, and continuously through the noise. Perhaps the spark ignited earlier still: strange tones heard long ago on a Radiomarelli valve radio, or the romantic notion of a lighthouse keeper sending a steady signal into the dark. When the project began to take form, I realised how deeply temperature, voltage and RF stability were intertwined. In a discussion with another friend - known simply as "Gippittì" - the final idea surfaced naturally: "Why not report everything by radio?" And from that moment, the beacon became two systems in one: HF for identity, VHF for self-awareness. The construction took months. Schematics were drafted, corrected, discarded, revived. Boards burned, rebuilt, tuned, and retuned. Code changed shape more than I care to admit. Many evenings vanished into the glow of a soldering iron and the hiss of a receiver. Mimmo, at one point, stopped hearing from me altogether. Only when the project finally neared completion did I send him a message:
"I can't wait to show it to you."
I know what he will say. He will smile and ask:
"And now what? Are you aiming for the Oscar?" And perhaps, yes - but this beacon lighting up the HF band was already enough.

When the RF section began, there was no "grand design". There was a Si5351 module, a handful of transistors, and a lot of trial-and-error. The first lessons were surprisingly basic: the difference between µF and nF, why a 100 Ω resistor is not the same as a 10 kΩ one, and why RF needs to be guided - almost tamed - instead of simply amplified.

The first stage was a simple buffer using classic "Handbook" favourites such as the 2N2222 or 2N3904. Its job was not power, but isolation: preventing the Si5351 from being pulled, distorted or mistuned by the following stages. It took several nights to understand why touching a single capacitor changed everything on the waterfall.

This section produced the first real RF you could feel under your fingers. Different devices were tested - BS170, IRF510, IRF520 - each one with personality, quirks and occasional smoke. The challenge was envelope stability: during CW keying, overshoot and ringing appeared everywhere until biasing, gate resistors and timing were tuned repeatedly.

The PA settled around a MOSFET running in a modest, well-behaved QRP configuration. The goal was not power - it was clean power. Every change in temperature or supply voltage shifted behaviour, and learning to stabilise it required INA219 sensors, thermal notes, and more than one burnt finger.

A multi-pole LC filter completed the chain. What looked simple on paper required manual retuning whenever the PA was modified. The final version delivers a clean 14 MHz signal with harmonics suppressed well below QRP guidelines.

This BEACON MACHINE is not a single board dressed as a project, but a small ecosystem where each module has a defined role and its own "personality". Removing one would break the balance of the entire system. The MASTER is the coordinator: it supervises timing, electrical safety, sensors, relay switching and the whole CW cycle. The OPERATOR is the HF unit: it generates the carrier, shapes the CW envelope and transmits only when the MASTER authorises the cycle. The REPORTER is the VHF APRS module: it receives telemetry from the MASTER, frames it in AX.25, modulates it in AFSK at 1200 bps and transmits it on VHF. Unlike the OPERATOR, the REPORTER is partly autonomous: it also sends its own periodic APRS beacons (position, status and identification) independently from HF activity. What may look like a secondary function is actually the backbone of the project. Nothing inside the RADIOFARO becomes "official" until it has travelled over the air. Built around an ESP32, an audio transformer, a simple NPN transistor for PTT and a DRA818V, the REPORTER captures the internal state of the beacon and gives it a voice. The MASTER condenses temperatures, currents, RF readings and cycle status into a compact message; the REPORTER turns that message into radio tones. The local digipeater then validates and forwards the packet. Only after the RF path is complete do APRS-IS and the Internet receive a copy - a delayed reflection of something already transmitted in radio. The OPERATOR and REPORTER therefore represent two faces of the same identity: HF says "who I am", VHF says "how I feel".

Through UART, that sentence reaches the REPORTER - still silent, still unmodulated - until the VHF stage shapes it into tones and turns it into a brief transmission. The local digipeater IZ6RND-12 hears the packet, validates it, and forwards it both on-air and into APRS-IS. Only after completing this journey does telemetry appear on the web. The Internet is not the source - merely the echo of something that has already existed in the spectrum. This VHF module is therefore not an accessory, but a reminder: the beacon depends on radio, and wishes to depend on nothing else. Inside the schematic, its simplicity is almost nostalgic: an audio transformer, a few capacitors, a pair of diodes, a transistor for PTT control, and a small amplifier for sidetone monitoring. It would look at home in a 1990s handbook - and that is part of its charm. The OPERATOR transmits the HF identity signal in CW, but only when authorised by the MASTER. The REPORTER handles the VHF side, generating AFSK telemetry packets and, when needed, its own APRS beacons. Together they form a dual voice: the OPERATOR delivers the beacon's RF presence, while the REPORTER broadcasts its internal state.

The beacon is, in essence, a practical RF experiment. Theory provided the starting point, but the real work happened under the soldering iron: measurements, corrections, thermal surprises, envelopes that refused to behave, and the usual "why is this drifting today?" moments familiar to anyone who has built RF circuits on a real bench. Every part of the system was refined under actual operating conditions. Rise time, harmonic content, bias stability and thermal behaviour were adjusted repeatedly until the CW envelope was clean and predictable. At the top of the architecture, three modules share the work:
• MASTER - the supervisory logic. It validates sensors, enforces safety, controls relays and cooling, and authorises every HF transmission.
• OPERATOR - the HF unit. It generates the carrier, shapes the CW envelope and transmits only with explicit MASTER approval.
• REPORTER - the VHF APRS module. It generates AFSK at 1200 bps, sends telemetry and, when needed, emits its own APRS beacons.
Around these modules lies the RF chain, managed entirely by the OPERATOR and monitored by the MASTER. A Si5351 clock generator provides a stable carrier, calibrated to within a few hertz. A small low-pass network directly after the oscillator reduces spurious content before the signal enters the RF chain.

Buffers and conditions the signal, preventing frequency pulling and setting the correct level for the driver. Lifts the signal to hundreds of milliwatts, shaping the RF envelope and maintaining stability during CW keying. The OPERATOR samples dBm here during the first seconds of every beacon cycle. Produces a clean QRP output (~1 W), monitored by INA sensors. Temperatures and currents are continuously supervised by the MASTER. A multi-pole LC filter tailored for 14 MHz ensures that only the intended RF energy leaves the system. It was tuned and adjusted several times, following the natural evolution of the PA. The OPERATOR bridges RF and logic, sampling sensors and shaping the RF envelope while the MASTER controls timing, safety margins and the overall transmission cycle. The REPORTER closes the loop by transmitting telemetry through VHF - the only authorised path to the outside world. In practical terms, the system behaves like a small three-unit team:

• the OPERATOR does the talking
• the MASTER decides when it is safe to speak
• the REPORTER keeps track of how the whole system feels
Provides the RF exit path for telemetry via APRS (AFSK 1200 bps). It is also capable of sending independent periodic APRS beacons (position + status), making it the semi-autonomous voice of the Radiofaro. No Wi-Fi, no Ethernet - only RF. A beacon built this way could operate indefinitely, needing only power. Add solar energy and a small battery, and it could live "forever," sending its quiet heartbeat into HF while reporting its condition in VHF.

Building IZ6RND/B was not only an electronic challenge. A beacon that thinks, measures, transmits and protects itself requires a significant amount of firmware - written, rewritten, patched, replaced, and sometimes completely thrown away and started anew.

The ESP32 platform is powerful but extremely peculiar: dual-core, asynchronous tasks, timing drift under load, hardware peripherals that behave differently from classic AVR microcontrollers. Managing CW envelope timing, UART traffic, AFSK modulation and safety logic inside a single cycle required countless iterations.

The project grew far beyond the early drafts. As of the current release, the codebase consists of:

• MASTER: ~2800 lines - Safety system, relay control, sensor validation, HF cycle authorisation, timing logic, UART protoco
• OPERATOR: ~3400 lines - CW keyer, envelope shaping, Si5351 control, ADC sampling, power measurement, coordination with MASTER
• REPORTER: ~1900 lines - APRS beacon generator, AFSK modem (1200 bps), AX.25 framing, periodic position/status beacons, UART telemetry reception

This does not include the modified libraries, local patches, test sketches, or debugging tools written along the way.

The following libraries were partially rewritten or patched to adapt them to the system:

• Adafruit_INA219 - Adapted for dual-bus configuration, corrected overflow behaviour, stabilised reading timing
• ESP32 hardware timers - Wrapped with a custom layer to ensure consistent CW envelope steps
• AFSK modem - Rewritten to avoid blocking loops and to maintain precise tone timing for APRS
• UART framed messaging - A lightweight protocol enabling MASTER -> REPORTER -> MASTER exchanges without desynchronisation

Each modification was tied to a practical necessity: preventing timing drift, avoiding audio artefacts in AFSK, or ensuring that telemetry wasn't lost during high-load moments.

Below are three real fragments extracted from the live firmware, illustrating the roles of each module

MASTER -> Handshake routine with OPERATOR

bool PingPong(String target, HardwareSerial &port, uint16_t timeout) { port.println("PING"); unsigned long start = millis(); while (millis() - start < timeout) { if (port.available()) { String resp = port.readStringUntil('\n'); if (resp == "PONG") return true; } } return false; }

OPERATOR -> CW envelope rise (click-free)

void envelopeRise() { for (int i = 0; i < 255; i += 5) { ledcWrite(0, i); delayMicroseconds(200); } }

REPORTER -> AFSK APRS modulation

void sendAFSK(uint8_t *data, int len) { for (int i = 0; i < len; i++) { afsk.modulateByte(data[i]); // AX.25 bitstream -> audio tones } }

If the RF chain gives IZ6RND/B its "voice", the firmware provides the "mind". Transmission timing, safety supervision, APRS gating and internal coordination are not hardcoded behaviours but the result of hundreds of small decisions encoded in firmware. The goal was not to write the smallest software possible, but to write software that behaves predictably under RF conditions. This became a rule early in the project: if a behaviour cannot be reproduced in RF, it does not belong in the firmware.

The RADIOFARO project grew over several months of continuous building, tuning, rewiring, rewriting, testing - and occasionally watching something burn on the bench. To give an idea of the scale, here is a concise snapshot of the journey.

• 2025-08-08 - First Si5351 module arrives (and dies during experiments)
• August - First soldering attempts on hand-made modules and first complete prototype assembled
• September - Experiments begin on CW envelope, modulation, AFSK
• October - Full focus on RF chain, PA stability, filters and thermal issues
• November - Programming marathon: MASTER, OPERATOR, REPORTER
• December - System integration: HF, VHF, APRS, telemetry, dashboard

• RF design and prototyping: ~300-350 hours
• Firmware development: ~450-550 hours
• Debugging, tests, measurements: ~250-300 hours
• Total estimated time: ~1000-1200 hours

• Si5351 modules destroyed: 2
• ESP32 modules temporarily "killed": 1 (miraculously revived)
• Various capacitors "sacrificed": several

• MASTER - 18 revisions, 2800+ lines
• OPERATOR - 24 revisions, 3400+ lines
• REPORTER - 14 revisions, 1900+ lines
• Combined commits (test sketches included): ~70-80 versions

• First detectable RF tone: 2025-08-25
• First clean CW envelope: early September
• First stable PA output (1 W QRP): October
• First APRS telemetry heard: mid-November
• Full HF/VHF dual beacon cycle operational: end of November

All of this was built on a small table, with a soldering iron that occasionally did more harm than good, and with the stubborn desire to make a beacon that was not only functional - but alive.

The Radiofaro communicates with two "voices":
HF (CW) - "Who I am"
VHF (APRS) - "How I feel"
Both are needed: one defines the signal, the other preserves the story of the machine.
This dual identity is the core of the project - and the reason why it works even when the Internet doesn't.
Why Everything Had To Be RF
Many beacons use Ethernet, Wi-Fi, or cloud dashboards.
Not this one.
The rule was simple:
If a piece of data has not travelled by RF, it does not officially exist.
This explains the presence of the REPORTER,
the UART bridge, and the AFSK chain - all designed so the beacon remains radio-native, not network-dependent.

---

Amateur radio has never been about flawless schematics. It is about putting energy into the spectrum and learning from the echo. IZ6RND/B is my first project of this complexity, and the fact that it works - cleanly, reliably, and autonomously - is already the reward. If you are reading this page, perhaps you have already heard the signal. Perhaps you were the first to reply. And if not, the beacon is still there, patient and steady, waiting for the next curious listener.

⚬ ARRL Experimental Methods in RF Design