Too long; Didn't read - hardware is far from marine grade; software is disgrace of software engineering.
Where do I start? From the beginning in 2015 when I bought the boat with full Simrad equipment, which was installed probably in 2004 by previous owner.
Main chart plotter inside had a faulty screen already, repeater in the cockpit stopped working 2 months later. All parts are discontinued and not repairable.
Next year flux-gate compass failed. After replacing there was no possibility to calibrate because of incompatibility with older controllers. All parts are discontinued. Moving to a newer Autopilot controller and some displays. Mounting sizes and holes are different size. Same year later - new chart plotter, radar and autopilot computer. Autopilot still maintains it's name - "an idiot".
In 2018 new windvane gets installed, lasts for 3 months and jams whole NMEA network, luckily after anchoring in a safe spot. Warranty replaces it. Also replacing NSS7 chart plotter, which cannot update itself anymore including some other software issues but warranty is overdue, need to pay extra, paying more for latest NSS7 evo3.
2021 - indirect lightning stroke while in the boatyard in St. Marys, Georgia. Luckily only electronics which were connected via NMEA were fried, though older devices from 2004 still had working displays. All devices got replaced to latest models except radar, hydraulic ram and rudder sensor and those have different connections. Once again, different mount points, need to cut off more plastic, drilling new holes.
Even after latest two NSS7 evo3 updates it crashes and gets into reboot loop. And It never remembers to store remind about updates setting, this issue is dragging for years now.
Next year the radar stopped working and after long debugging with the support - failure inside radar, warranty is overdue, so no replacement.
Also original HLD350 MK2 Hydraulic starts leaking, no service parts available after contacting support. Ordered and installed a new T1 hydraulic ram. On the way from Costa Rica even on closed haul course it disengaged autopilot multiple times because of "no rudder response". After two weeks of calibrating and corresponding with support and two overnight sails between Galapagos islands I eventually left towards Pitcairn island for about 3 weeks passage. Outcome was outrageous.
Sometimes it disengaged 10 times an hour. Emailing with support via satellite phone was useless. The rest down bellow was nearly impossible and sometimes I would leave only the main sail on 3rd reef so I just could have a quick rest. After nearly 2 weeks I finally gave up, hove-to and installed back the old HLD350 MK2 and refilled hydraulic fluid. Things got way better but not perfect as autopilot still would disengage occasionally. The conclusion - the autopilot software was written by idiots and it felt like it does not use data from precision compass which has turn/roll/pitch data (heave data seems to be faulty).
The limit of me and the boat was when I had to run downwind towards Gambier islands from a low pressure zone while blowing 35 knots and having waves from two sides. I could not rest inside as it was raining all the time and it would take too much time to get up and steer back the helm and switch back the autopilot. So I had to nap on hard corners of the cockpit with full gear near the controller so I could press the auto button whenever I heard a beep.
It was time to do something about it. To replace all hardware with more capable brand or to write a better software. All data was already received via PICAN-M hat and Raspberry Pi so it was the matter of sending the same data the autopilot controller does when you press "auto" button whenever the autopilot computer does not get a response from the rudder. What a shithole (rabbit hole is a too nice description) I had to get into! Many NMEA2000 Parameter Group Numbers (PGN) were used for the multiple purposes and you would need to collect data from separate pgn's for the same intent. Even some sentences were just encoded text like NMEA0183 into binary... Opensource projects like Signal K and canboat were not able to handle such nonsense so had to write my own reader and parser.
After sailing more than 2500 nautical miles the result was quite satisfying as I am able to properly rest as I hear only a short beep when autopilot disengages and engages and steers back to the saved mode and course. Also now I have a remote controller in my laptop and smartphone.
The only issue that it still steers like an idiot. It just needs to turn one second earlier as humans do and the problem would be solved. Or just to track rate of turn properly on not to oversteer when the boat is pushed by secondary swell.
The next step would be the hardware and software combinded and the pypilot computer is already onboard. Just wandering if I want to get into another headache and to make something functional, or maybe it might be fun.
Navico - "The most trusted brands in the industry" they say...