Tern · project
The line's gone, or you got close but 15ft visibility doesn't make close the same as close enough. This is meant to tell you which way to swing.
An idea in preliminary design; it should be possible, but isn't done yet.
What it is
Sometimes the line is down, or you can get close to a wreck but can’t quite find it. The idea is simple: mount it on the front of the scooter, hold it level, and swing slowly through a circle. With five transducers, you get some indication as the array swings onto a target, confirmation as it sweeps through the next one, and a highlight when you’re pointed straight at it — enough to say “generally this direction.”
What it isn’t
Military and professional packages can pick out submillimeter detail hundreds of meters out. This will not be that. You can't do that in this price range. You can't come close. The transducers under consideration have a 15–20° beam width, so this is a coarse pointer, not a bearing fix — and right now it’s earlier than that: the board is ordered, not assembled, and nothing has been tested in water yet. I've done quite a bit of ultrasonic work, so I'm confident I can build something that works. The question will be whether it works well enough to make it a practical tool for actual divers.
Where this stands
The array is the whole point, but it’s not where this starts. Phase 1 proves the electronics on one channel before any cost or complexity goes into five of them.
One TUSS4470 front end driving one 235 kHz piezo disc, on an Adafruit Feather ESP32 V2. The question this phase answers: can the electronics generate a clean burst, read a believable echo envelope, and tell a real bottom return from air and noise? Firmware is written; waiting on board and parts to arrive.
Five elements, a shared low-cost driver, and a 12–24V pack in a 3″ Blue Robotics housing. This is the actual concept — direction from sequential returns across a sweep — and it only gets built once Phase 1 proves the single-channel chain works. One huge challenge will be building a modular array of transducers that actually holds out water, and isn't radically complicated, and can be built by a hobbyist.
Nothing here is new to me, but the water is. I’ve done ultrasonic work before and have a handle on how the signal chain should behave —. But confidence isn’t data, and until the first housing is wet-outside-dry-inside and providing useful data, every number on this page is a design target, not a measurement.
Specs
Phase 1 hardware and firmware, plus the Phase 3 target this is all in service of.
| Front end | TI TUSS4470 ultrasonic AFE — burst generation, H-bridge drive, bandpass filter and echo-envelope capture in one chip. |
|---|---|
| Transducer (Phase 1) | STEMINC SMD25T85F234S, a single 235 kHz piezo disc — a known part, chosen to validate the electronics before committing to an array. |
| Processing | Adafruit Feather ESP32 V2. Generates the burst on the RMT peripheral at 12.5 ns resolution and samples the echo envelope on an onboard ADC. |
| Interfaces | A serial CLI and a web UI at sonar.local share one ping engine and return identical JSON; the web side adds a live chart, auto-ping and CSV download. |
| Range axis | Switches speed-of-sound between air (343 m/s) and water (1480 m/s). Near-range numbers currently read slightly short — burst start time is taken in firmware rather than triggered off the real first pulse — until it’s calibrated against a known distance. |
| Planned array (Phase 3) | Five elements in a 3″ Blue Robotics housing, a low-cost shared driver, and a 12–24V battery pack. Target cost is under $1000 per unit, possibly as low as $500. |
Roadmap
Directional, not promised — more than most Tern projects, since none of it is tested yet. The further right, the more likely it changes.
Actively being built or tested
Committed, not started
Directional, may change
Open problem. Whether a 15–20° beam width across five elements gives useful left/right/centre discrimination once it’s actually in the water. Nothing has gotten wet yet, so this is a bet backed by prior ultrasonic experience, not a measured result.
Source & docs
The source isn't public yet. It opens when Sonar is far enough along that you could genuinely copy the work — a repository you can't build from is just a pile of files. What it's waiting on: something worth publishing — this one is early enough that there is no repo yet.
Licensing
Undesignated
Source-available, not OSI-open: free for divers and tinkerers, not for commercial resale. The full position — including the labour-for-a-friend carve-out and the Tern trademark note — is still being written up.
Nothing about this project is validated yet — not the range accuracy, not the beam-width discrimination, none of it. Even once it is: a coarse direction indicator is not a reason to abandon standard lost-line or lost-diver procedures, and it is not a substitute for training, a reel, or your own judgement.
This is a hobbyist, DIY instrument with no warranty and no certification of any kind. Read the full safety position →
