Outboard Market: Consolidated and Closed

▶ Watch (0:25)

The five major outboard manufacturers above 150 horsepower are Mercury (owned by Brunswick), Yamaha, Honda, and Suzuki. Mercury sent Lorman a letter citing legal liability concerns. Yamaha’s integrated steering motor locks buyers into Yamaha’s helm exclusively. Station-keeping features like Virtual Anchor and Skyhook are black boxes. Mercury won’t disclose how they work, even to Garmin. The reason isn’t technical. It’s warranty exposure and not wanting autonomous boats to kill people accidentally.

Jet Skis and Cable-Driven Systems: Read the Voltages

▶ Watch (7:10)

The simplest entry point is jet skis. The CDU wiring diagram, available in the service manual, shows the throttle uses two Hall effect sensors outputting ratiometric voltages. The ECU reads those voltages and sends CAN or analog to the throttle bodies. To control the engine, replicate those voltages with a DAC. Slew rates matter. Match what a human hand would do. Jet skis are designed for beginners, so the ECU won’t error out on imprecise inputs, but clean voltage ramps are safer.

Mercury DTS: Three Layers Between Helm and Engine

▶ Watch (9:33)

Mercury’s DTS architecture separates into three layers. Engine space: each ECU manages firing sequence, timing, water pumps, and thermal shutdown independently. Boat space: command modules know the full motor configuration, including port, starboard, and center engines. User space: the helm translates intent to go forward into per-engine commands. DTS1, pre-2022, uses analog helm signals. DTS2 moved to CAN from the helm. Either way, user space sends three voltages per throttle handle. Plus 100 is full forward, zero is neutral, minus 100 is full reverse.

Glendining: Proprietary CAN, One Afternoon

▶ Watch (12:51)

Glendining Controls uses a proprietary CAN bus, not J1939, so standard marine decode tools fail immediately. The real interface is simpler. Open the helm and it’s a Hall effect sensor on a magnet. The lever moves the magnet; the sensor outputs a voltage. Figuring this out required a two-channel oscilloscope. A multimeter lacked the impedance. The full reverse-engineering effort, including a lunch break, took one afternoon. CP modules handle multi-engine coordination and come preconfigured from Glendining if you need them.

Pretend to Be the Helm

▶ Watch (17:34)

PLCs failed. MQTT message rates from Automation Direct’s STM32 hit undocumented limits. The working solution: a Teensy with peripheral hardware listening to PWM hobby servo signals or Mavlink over serial, then outputting DAC voltages to the engine’s user-space connector. About 500 lines of code. The principle: never reverse-engineer Mercury’s internal CAN buses. Those buses will change under you. Intervene at the user-space level, the helm connector, and the engine handles everything below. Use signed 8-bit integers to match Mercury’s -100 to +100 range.

Q&A

Will the industry eventually drop cable-driven systems? Yamaha is heading that way, and new digital wiring harnesses already cost over $1,000 to rig with no self-service option. ▶ 20:30

Why not reverse-engineer Mercury’s CAN bus directly? Mercury likely uses non-standard binary encoding that takes far more than an afternoon to decode, and the buses will change under you anyway. ▶ 21:29

Can the Maritime Hacking Village badge implement this approach? Yes, it supports NMEA 0183 and NMEA 2000 and has arbitrary voltage outputs. ▶ 22:21

Notable Quotes

so all that goes out the window. Alex Lorman · ▶ 1:22

honestly, jet skis are really easy Alex Lorman · ▶ 9:13

You just ask them nicely. Alex Lorman · ▶ 14:56

works. It works reliably. Alex Lorman · ▶ 19:34

Key Takeaways

  • Intervene at the helm’s analog voltage output, not Mercury’s proprietary CAN buses.
  • Hall effect sensors and ratiometric voltages govern throttle control on every major outboard system.
  • A Teensy reading Mavlink over serial and outputting DAC voltages controls real engines in 500 lines of code.

About the Speaker

Alex Lorman grew up in Washington, D.C. and earned a B.S. in Architecture from Catholic University of America. His work on oil and gas projects and marine salvage led him to conclude that the maritime industry needed robotics. He co-founded Sea Machines in 2014 and moved to the Boston area to lead the effort. He enjoys working on cars, ships, bicycles, and anything with a mechanical or electrical heart.