Projects to come.

The list

Four big ones, then the smaller jobs behind them

1. LoRa mesh messaging

An experiment: a free, private radio mesh between boats and people

Every node passes the message to the next. No tower, no provider, no account. A message from Boat A can reach the campfire by hopping across Boats B and C.

Talk to the rest of the anchorage, or to a friend over the hill, with no cell service, no internet provider and no account anywhere.

Why we want to try it

Straight answer: it is not filling a gap on this boat. We already have VHF, cell service and Starlink. This is an experiment, an interesting hobby, and a way to find out what it leads to. Still, a few reasons it is worth the afternoon:

• A second, independent radio. Lose the VHF antenna — a mast hit by lightning, a masthead antenna that comes down — and the VHF is close to useless. A small LoRa node has its own antenna and its own battery, so it keeps working when the VHF cannot. At best that is a way to message boats nearby. It is not a distress system and does not replace a handheld VHF, an EPIRB or a personal locator beacon.
• Talking to the boat's bot from shore with no signal. If you are already carrying a phone ashore, a pocket node paired to it could send a message over the mesh to a node aboard, which hands it to the Pi and the Telegram bot. With cell coverage, Telegram already does this and the mesh adds nothing. It is the no-coverage case that is new. We think it can be done, and that is something to test rather than promise.
• New ways to communicate. Nobody knows yet what a cheap, free, private radio mesh turns into once boats and people start using it. That curiosity is most of the reason.

What LoRa is

LoRa is a radio modulation built for sending very small amounts of data a very long way on very little power. It runs in the licence-free band — around 915 MHz in the US — and it trades speed for range in a way no WiFi or Bluetooth radio does. Think text messages and a position dot, not photos and not voice.

The two programs that turn the radios into a mesh

We have not picked. It may come down to which one the people around us are already running, because a mesh with one node on it is just a very slow radio. That is the whole game: it only gets good when more people join in. The same participation argument as the crowdsourcing section. Links: Meshtastic and MeshCore.

How it would work between boats

• In an anchorage. Twenty boats in a bay, all within a couple of miles, every one of them a relay. That is the exact shape of network this was designed for. Dinghy-to-boat chat, “who's going ashore?”, dragging-anchor warnings, all for free.
• Offshore, in company. Be honest about this one: LoRa will not cross an ocean. Every hop is a line-of-sight radio path, so a boat reaches the boats within some miles of it, not the shore and not the other side. What it gives a group of boats on a passage is a free group chat and position sharing between each other, with no satellite airtime.
• Over water is the best case for radio. No hills, no buildings, no trees. A node on a masthead antenna is about as good as a line-of-sight path gets.
• It costs nothing to run. No airtime, no plan, no provider that can go out of business. A node draws so little that a small solar panel runs one indefinitely.
• After a storm. When the cell towers are down and the marina WiFi is off, this is unaffected. It never depended on any of it.

Away from the boat: the campfire case

The same radios work on land, which is where it gets interesting for everybody, not only sailors. Camping, hiking, hunting, a day on the beach, a work crew in a valley — anywhere there is no cell service. If everybody in the group carries a small node, they can text each other and see each other on a map, and a node on a hill or on a boat's mast carries the signal further for everyone.

A note on phones. A phone has no LoRa radio in it. The usual setup is a pocket-sized node with its own battery that pairs to the phone over Bluetooth, and the phone becomes the keyboard and screen. Some nodes have their own little screen and keyboard and need no phone at all.

Typical range between hand-helds in woods or between buildings is short, perhaps a mile or two. Higher ground and open water are much better. Those are rules of thumb, not promises, and they depend on the antenna, the terrain and how many other nodes are about.

What we would build

1. A node aboard with a screen, so the simplest case needs no phone, no app and no pairing. Text and position on a little display at the nav station.
2. Position sharing with the boats around us, which in an anchorage is useful and in company offshore is more so.
3. An MQTT bridge. Meshtastic can push messages to and from MQTT, the standard language everything else on this boat already speaks. That turns the mesh into one more input rather than a separate island.
4. Bridge it to the Telegram bot. A message arrives on the mesh from a boat a quarter mile away, goes into MQTT, and reaches a phone. Or the other way round.
5. And to the onboard AI. Ask a question over the mesh, have the boat's offline model answer, and send a short reply back. No internet anywhere in that chain, which is either very useful or very silly, and we want to find out which.
6. Alarms out. A bilge alarm or an anchor drag alert reaching your phone while you are ashore in a dinghy with no coverage.

2. Our own sonar

Open Echo: owning the whole chain, from the ping to the pixel

No proprietary transducer, no proprietary box.

Sonar is the last genuinely closed thing on this boat. We decoded the fishfinder's output, which means we can read what the Axiom produces — but we are still asking a proprietary box to do the actual work. Switch the MFD off and there is no depth.

The answer is Open Echo: open-source sonar hardware, with schematics and board files published and raw echo data out the other end. A TUSS4470 driver board that will fire a real transducer and digitise what comes back.

The chain we want to own:

1. Drive our own transducer. Our own pulse, our own frequency, our own timing — not whatever a sealed box decided.
2. Capture the raw return. Echo strength against time, before anybody has processed it into a number.
3. Do our own processing. Bottom detection, gain and the waterfall, in code we can read and change.
4. Draw it on our own screen. Which already works. It just needs feeding from our own hardware instead of theirs.

Open Echo's tested transducer list includes Raymarine CPT-S 50/200 kHz units, so the transducers already fitted to boats like this one are not the lock-in. The box is. Depth that doesn't depend on a display being powered is the point on a boat at anchor.

3. A digital Ford Lehman

Putting the engine's gauges on the network without buying a gateway

Oil pressure, temperature, RPM and hours, logged and trended instead of glanced at. The engine itself →

The Ford Lehman 120 is a simple, mechanical diesel with analogue gauges. There is no engine computer to plug into. The commercial answer is a box that reads the old senders and puts the numbers onto NMEA 2000. Actisense makes one, the EMU-1, which currently sells for roughly $600 to $720 at US marine retailers, and other makers sell similar converters. It is good, sealed, tested equipment, and it costs what it costs because marine is a tiny market.

The plan is to do the same job with Raspberry Pi parts and do more with the result:

• Read the same senders the gauges use. Oil pressure and coolant temperature senders are analogue, so a small analogue-to-digital board next to the Pi reads them. Our estimate is tens of dollars in parts.
• RPM from a pulse. Count pulses from the alternator or the existing tachometer signal, and engine hours fall out of counting the time it is running.
• Straight into Signal K. Which means the numbers appear on every screen aboard, and onto NMEA 2000 too if we want them on the plotter. We already do this for the generator.
• Better, not just cheaper. Logged history, a trend for each reading, hours feeding the maintenance list in the ship's log, and an alarm to your phone when oil pressure sags, rather than a needle you have to be looking at.

4. Weather Now

Instant local weather, from sensors that already exist

Watch a squall coming up the bay, because somebody five miles down it is already being hit by it.

A forecast tells you about this afternoon over a 30-mile grid square. It cannot tell you there is a squall line eight miles west of you right now, moving this way, with 40 knots in front of it. That is the information that matters when you are deciding whether to reef or run for cover, and almost nothing serves it to a boat.

But the data exists. Scattered across the water and shoreline around you there are hundreds of weather stations — on houses, on docks, on masts of other boats — many of them publishing live readings to public networks. Plus NOAA buoys, plus airport observations, plus live lightning data. And that is why crowdsourcing matters here more than anywhere: every extra boat that shares its masthead instrument is one more point on the map, and the picture gets sharper with each one.

What we would build

• Pull every open station within a useful radius of the boat's own position, automatically, as the boat moves. See what somebody's masthead wind instrument is doing five miles down the bay.
• Watch them as a field rather than one at a time. Pressure falling at three stations upwind in sequence is a system arriving, and it is visible long before it reaches you.
• Squall detection. A sharp pressure jump and a wind shift at a station upwind is a gust front. Work out the bearing and speed and you get an arrival time.
• An app that writes the trend in plain English. Something like “pressure falling at three stations to the west, wind backing and building, front roughly forty minutes away.” That line is an illustration, not real output.
• Fold in the boat's own instruments — masthead wind, barometer and the sea state sensor — so the boat is one more station in its own network.
• On top of Windy and PredictWind, not instead of them. Those are forecasts. This is what is happening now. They answer different questions and each makes the other more useful.

Nowcasting rather than forecasting: not what the model thinks about tomorrow, but what is happening within twenty miles in the next hour. And offshore, where there are no stations, it inverts: boats become the network, sharing real observations with each other instead of all guessing from the same GRIB.

And the rest of the list

Smaller jobs behind the big four

What's already built → Open vs closed