Sailing Auto Pilot - Competition - Ready For Tomorrow 2022

Who m I: a nerd with passion in sailing that fixes marine electronics reverse engineering most of them since schematics are not available. Electronicmarine on youtube.

Premises: An autopilot system used on a mini, especially when solo sailing must fulfill a serries of requirements:

TO BE RELIABLE FIRST OF ALL!!! (an unwanted chinese jibe when you are on the deck can kill you in the middle of the ocean)

To draw less power possible, since the amount of energy onboard is really limited

To be able to steer the boat under heavy wind and sea conditions

Last but not list: to keep the boat fast under all range of wind and sea- after all is a race!

Start with the basics: how an autopilot for a boat work: the compass module (sometimes called fluxgate) take a reading, compare it with a set course, then command the actuator to move the helm to try to reach that bearing, the helm angle is read by a rudder reference sensor to know how much it moves then corrects until the desidered course is reached (in theory) in practice things complicate a lot more (maybe will go back to this argument later- ask me how I know :) ).

The second operation mode desired on a mini is have it operate on the TWA (true wind angle), and here the things complicate a lot more, first of all you need an acurate wind sensor with dedicated wind speed data and wind angle data, fast data, possibly with the posibility to change the damping and averaging of readings depending on wind and sea conditions. Altough is feaseble to buid one, it's much simpler to get an already made solution on the market since the adc converssions, callibrations, various options in dealing and processing all that part will take a lot of space and processor working time and just makes debughing and proggraming a lot more difficult and it requires aeither a separate processor or a multitasking real time one that is not easy to proggram by an average arduino user. After having a relible wind data, it needs a mems packege (gyro and accelometers) barometric pressure or water temperature, or air temperature not really needed since is not a NASA prodgect who take in consideration all those variables, Focus on the important things first. Having solved the reading of the mems packege and sucessfully implemented a kalmann filter the rest of the math can be solved knowing the mast height (they are in class requirements) and calculating the position of the mems location from the center of rotation of the boat you can then calculate the TWA and TWS when the boat gets leaned and throwed away by the waves.

Having solved that part the rest can be easy (or at list is seems) moving an actuator motor and get back the feedback from a rudder sensor.

The rudder sensor used by a lot of manufacturers is relativeley simple, I can even 3dprint them, one potentiometer enclosed in a waterthight enclosure, they use quality vishay or burns industrial potentiometers with values between 2.5 and 5k.

The actuator arm and system is what stayed behind in the past 30 years, there ater basically 2 systems: one linear ballscew type and one with an electric reversible hydraulic pump that moves a piston, (raymarine t1/st4000 tiller and the second type lecomble&scmidt or the ramdrive from b&g are the most used on mini) but both have one base problem: ineficiency in transforming the energy into movment due to the classical brushed motors used inside. In the past 15 years they have tried to minimise the power draw by using pwm to controll a soft start and operation of the dc motros.

PROBLEM WITH ALMOST ALL AMATEUR BUILD AUTOPILOTS (and some commercial ones also): most of them do not incorporate a monitoring system for what is the power draw of the motor and do not incorporate protections for motor unwanted stall or if is forced, this blow up the mosfets in the H brige!!!! so in order to have a certain reliabilty on the system a somekind of amperometric protection must be designed from the beginning, it might seem easy to implement but since is pwm controlled it can be a challenge for an untried eye (just the timing of the reading and interrupt routines can be a bit tricky). For that purpouse the hall sensors that are prone to interferrences do not work well in this kind of the applications and a shunt solution is much preffered.

Since i've mentioned the interferrences, most hobysts just think that the power in a boat is clean as in a lab, definetley not true, hence all analog imputs must be properly filtered and protected against spikes, ripples and other type of interferences present onboard, agaian ask me how I know :) The adc pins that read the rudder sensor for instance and adc reffrece voltage on 5v microcontrollers for example is a lot better to be used on a 2.54v reffrence source, diode clamped externaly and pass thru resistor divider and then again a protection fast zener used... inductance filters also welcome on those inputs :)

Now building it for the cheap on the hardware side is relatively easy, the part that is the most expensive is proggramming and debbughing...

here you have 2 options: go with a lot of power like using a PI, or break the system in multiple microcontrollers and designate and optimise speciffic tasks with interrupt driven routines in order of importance and needs.

I2c communications for amateur level in this kind of projects is better to let them aside if you are not custom build the pcb and take care in path design, altoug a lot of times they work, or seem to work, in alt more cases make bad jokes, intermitent failures that make debugghing a nightmare or induce catastrofical failures on the micro side...

power filtering is a really important aspect also, automotive grade components are recomanded in this cases- cheap from car junkyard, a bit difficult to desolder but not that hard.

Now for the display part: thiese days everybody wants nice colors led displays that have inside powerfull drivers and display nice grphics, the problem is that they consume too much power, so you can use one for debughing purpouses while you are building the system, but for the race time a ultra low power display with efficient back lightning is most recomanded.

Since i've mentioned the low power: the micro-controllers you use should be possibly selected in the nanowatt family or a real ultra low power microcontrollers, the sleep routines should be used and wake up interrupts on speciffic events since you don't need it to run all the times...

lLEARN FROM MY MISTAKES:

do not use an H bridge Ic, have it custom build with discrete power mosfets and have in place an amperometric protection that autoresets itself.

Use optocouplers between the micro-controller and power electronics, less magic smoke will be released :)

properly filter the power parts and separate the power rails

then everything else can be at later date integrated with an Bluetooth or wifi module in order to be send on some phone or tablet, although the class limitations do not allow them.

At this point a way to easy remotely program the logic with a laptop can be maybe a lot more useful feature instead of inserting usb or ICSP

Who m I: a nerd with passion in sailing that fixes marine electronics reverse engineering most of them since schematics are not available. Electronicmarine on youtube.

Premises: An autopilot system used on a mini, especially when solo sailing must fulfill a serries of requirements:

TO BE RELIABLE FIRST OF ALL!!! (an unwanted chinese jibe when you are on the deck can kill you in the middle of the ocean)

To draw less power possible, since the amount of energy onboard is really limited

To be able to steer the boat under heavy wind and sea conditions

Last but not list: to keep the boat fast under all range of wind and sea- after all is a race!

Start with the basics: how an autopilot for a boat work: the compass module (sometimes called fluxgate) take a reading, compare it with a set course, then command the actuator to move the helm to try to reach that bearing, the helm angle is read by a rudder reference sensor to know how much it moves then corrects until the desidered course is reached (in theory) in practice things complicate a lot more (maybe will go back to this argument later- ask me how I know :) ).

The second operation mode desired on a mini is have it operate on the TWA (true wind angle), and here the things complicate a lot more, first of all you need an acurate wind sensor with dedicated wind speed data and wind angle data, fast data, possibly with the posibility to change the damping and averaging of readings depending on wind and sea conditions. Altough is feaseble to buid one, it's much simpler to get an already made solution on the market since the adc converssions, callibrations, various options in dealing and processing all that part will take a lot of space and processor working time and just makes debughing and proggraming a lot more difficult and it requires aeither a separate processor or a multitasking real time one that is not easy to proggram by an average arduino user. After having a relible wind data, it needs a mems packege (gyro and accelometers) barometric pressure or water temperature, or air temperature not really needed since is not a NASA prodgect who take in consideration all those variables, Focus on the important things first. Having solved the reading of the mems packege and sucessfully implemented a kalmann filter the rest of the math can be solved knowing the mast height (they are in class requirements) and calculating the position of the mems location from the center of rotation of the boat you can then calculate the TWA and TWS when the boat gets leaned and throwed away by the waves.

Having solved that part the rest can be easy (or at list is seems) moving an actuator motor and get back the feedback from a rudder sensor.

The rudder sensor used by a lot of manufacturers is relativeley simple, I can even 3dprint them, one potentiometer enclosed in a waterthight enclosure, they use quality vishay or burns industrial potentiometers with values between 2.5 and 5k.

The actuator arm and system is what stayed behind in the past 30 years, there ater basically 2 systems: one linear ballscew type and one with an electric reversible hydraulic pump that moves a piston, (raymarine t1/st4000 tiller and the second type lecomble&scmidt or the ramdrive from b&g are the most used on mini) but both have one base problem: ineficiency in transforming the energy into movment due to the classical brushed motors used inside. In the past 15 years they have tried to minimise the power draw by using pwm to controll a soft start and operation of the dc motros.

PROBLEM WITH ALMOST ALL AMATEUR BUILD AUTOPILOTS (and some commercial ones also): most of them do not incorporate a monitoring system for what is the power draw of the motor and do not incorporate protections for motor unwanted stall or if is forced, this blow up the mosfets in the H brige!!!! so in order to have a certain reliabilty on the system a somekind of amperometric protection must be designed from the beginning, it might seem easy to implement but since is pwm controlled it can be a challenge for an untried eye (just the timing of the reading and interrupt routines can be a bit tricky). For that purpouse the hall sensors that are prone to interferrences do not work well in this kind of the applications and a shunt solution is much preffered.

Since i've mentioned the interferrences, most hobysts just think that the power in a boat is clean as in a lab, definetley not true, hence all analog imputs must be properly filtered and protected against spikes, ripples and other type of interferences present onboard, agaian ask me how I know :) The adc pins that read the rudder sensor for instance and adc reffrece voltage on 5v microcontrollers for example is a lot better to be used on a 2.54v reffrence source, diode clamped externaly and pass thru resistor divider and then again a protection fast zener used... inductance filters also welcome on those inputs :)

Now building it for the cheap on the hardware side is relatively easy, the part that is the most expensive is proggramming and debbughing...

here you have 2 options: go with a lot of power like using a PI, or break the system in multiple microcontrollers and designate and optimise speciffic tasks with interrupt driven routines in order of importance and needs.

I2c communications for amateur level in this kind of projects is better to let them aside if you are not custom build the pcb and take care in path design, altoug a lot of times they work, or seem to work, in alt more cases make bad jokes, intermitent failures that make debugghing a nightmare or induce catastrofical failures on the micro side...

power filtering is a really important aspect also, automotive grade components are recomanded in this cases- cheap from car junkyard, a bit difficult to desolder but not that hard.

Now for the display part: thiese days everybody wants nice colors led displays that have inside powerfull drivers and display nice grphics, the problem is that they consume too much power, so you can use one for debughing purpouses while you are building the system, but for the race time a ultra low power display with efficient back lightning is most recomanded.

Since i've mentioned the low power: the micro-controllers you use should be possibly selected in the nanowatt family or a real ultra low power microcontrollers, the sleep routines should be used and wake up interrupts on speciffic events since you don't need it to run all the times...

lLEARN FROM MY MISTAKES:

do not use an H bridge Ic, have it custom build with discrete power mosfets and have in place an amperometric protection that autoresets itself.

Use optocouplers between the micro-controller and power electronics, less magic smoke will be released :)

properly filter the power parts and separate the power rails

then everything else can be at later date integrated with an Bluetooth or wifi module in order to be send on some phone or tablet, although the class limitations do not allow them.

At this point a way to easy remotely program the logic with a laptop can be maybe a lot more useful feature instead of inserting usb or ICSP

Now broken into pieces and maybe with the comunity help maybe wee can really buid a working cheap syetem, the wind part for the mechanics 3dprinted, rudder sensor and power filtering parts can be build easy... the hardware design and phisically build is easy to me also to do in really fast timeframe since i stihave all my protoiping tools in house... the proggraming and debughing time is the one i don't have...

Congratulations riky_electronicmarine This comment is the winning comment! I have sent you a message regards your prizes!