homemade autopilot

[X7JAY7X]

zorbas. I looked for a long time and could not find one. I had a few ideas on how to make my own clutch to use with a linear actuator but I ended up going with a steering box instead with integrated clutch. You may want to look at a hydraulic solution. Some sort of a bypass on the hydraulic cylinder may be your best bet.

[Fanie]

Originally Posted by zorbas

Hello,

I have been planing to build my own autopilot for a while now. Mostly because I think it would be fun. I have everything more or less sorted except for the drive system. Since it will be operating directly on the quadrant below deck I need to find a linear actuator that has a clutch. But I simply can't find one.. any ideas?

Wiper motor with a rack & pinnion like the gates use. I think a push pull cable to your steered component would be best.

The position encoder can be a bunch of resistors in series with reed switches between each resistor. So as the magnet slides back and forth over the switches the on switces will output a voltage proportional to the position. You can add as many as you want, and waterproof the encoder by potting it. I made some and it works great, if you use 101 resistors and 100 reeds you have a 1% encoder...

[Fanie]

So if you have a wind vein with a pot in it and the position encoder as above you can use a comparator to control the motor left or right. To drive the motor you need a H-bridge - attached.

I use a micro to do the controlling with but an analog circuit will do as well. You have to provide 'dead time' between swirching motor direction or the FET's will short circuit the supply - which isn't fun ! Easy to do with a micro but as I said can be done with analog also.

[Fanie]

This is what the encoder looks like. I had PCB's made and use SMD resistors to increase resolution. The long position encoder is 400mm long for rudder steering (or outboard) and the smaller one is 100mm for outboard throttle or gear position control.

The slots in the PCB is where the reed switches fit in for easy soldering to the PCB. The reistors sit on the side of the PCB, I use 1206 resistors. The encoder's current can be calculated by counting the amount of resistors and devide the 5V supply by it. ie 5V / 100 x 1k = 5mA. The lower the resistors the higher the current, and vice versa. 5mA to 15mA should be good enough.

The housings are thin SS sheet I laser cut with holes for mounting, folded, put PCB in and pot water proof.

[Fanie]

This is a simple drive I made before. It uses one motherly big 24V bus wiper motor, even at 12V I cannot physically stop the drive when pushing or pulling, so it is strong.

The encoder will simply mount along the moving arm and a magnet gets fitted to the arm. It will output it's position as the magnet moves with the arm.

[X7JAY7X]

Fanie, how do you have a clutch with that?

Why did you make a linear encoder over buying one? They can be expensive but if you look hard enough you can find one for a good price.

[Fanie]

Originally Posted by X7JAY7X

Fanie, how do you have a clutch with that?

Why did you make a linear encoder over buying one? They can be expensive but if you look hard enough you can find one for a good price.

Making the encoder wasn't THAT cheap - in SA everything is a ripoff

Also, I kinda design and manufacture electronic things I can however make my encoders any size I want - even meters long if it comes to that. Some of my friends also wanted some custom ones and I made them without problems.

As for a clutch, depends how much movement and how rigit you want it. A push-pull spring can perhapa work with a spring centering the 'cluth'. The clutch is then two sleeves (pipes) one sliding into the other and the spring keeps the two parts centered. The advantage would be that the position right itself in case your rudder gets hooked on a submerged container

There may be other ways of doing it also - one would be a force sensor (like in a load cell) that would, when the push or pull force exceeds a certain amount, reverse the motor or rather rotate it in the direction of force to prevent breakage.

[X7JAY7X]

The clutch is then two sleeves (pipes) one sliding into the other and the spring keeps the two parts centered.

This was one of my clutch ideas but instead of the spring, the pipes would lock together. I don't like the idea of the spring because you will be fighting the spring when you want to steer the boat.

[Fanie]

A little spring movement will have a shock absorber effect. The spring should be stiff enough to be affected only slightly under normal conditions. I think the drawback is the size rather than the working.

I also ditched the idea and am looking at flip up rudders instead, perhaps make it so they can be replaced more easily. Haven't got any final ideas yet but it's in the back of my mind (somewhere...).

[zorbas]

It is so strange that there simply doen't seem to be a linear drive with a built in clutch.

Maybe I'll have to install a cogwheel on the quadrant and make it chain driven. Then it would be easy to install a electromechanical clutch. Or install a clutch on the pole itself and connect the cogwheel to the clutch. How to get the clutch onto the steering pole is of course not tthat easy

[Fanie]

It can be done rather easy - same as a fishing reel clutch.

You need a shaft with thread on the end with a flat surface on the motor and the shaft protrudes through the flat surface.

Another flat surface on the sprocket with a center hole.
Add one or two friction disks also with center holes that goes between the two flat surfaces.

Washers with nylock to set the friction you need for slippage.

The left orange washer is attached to the shaft
Then 3 clutches,
then an orange washer attached to the big blue sprocket,
the two washers
and a nylock.

[breezetrees]

Hi everyone,
I've enjoyed this thread. I made an autopilot for my boat too, here's a link to the page that describes it:

http://www.holdentechnology.com/comp...made-autopilot

It uses an inertial measurement unit/rate gyro/compass that I got from some old consulting work, an atmel 8 bit processor (programmed in C), a drive unit from a broken autopilot, and a motor controller designed for hobby robotics.

I housed the electronics in an old LORAN unit case because it is waterproof and has an LCD and keypad. The control law is full state feedback but with the states I have it's very similar to PID. I have a background in autonomous systems but this was a fun project.

I recently added a datalogger, there is some data on the website, if anyone has something to compare I'd love to see it. I think I can get some improvements but it's working good so far.

-Mike

[frasco]

Wow, that's awesome! I also have an old Plastimo lying around (with a broken compass unit) and I intend to use it in my project. Though I'm still not too eager to start building yet as there are many months of snow and ice ahead but this will surely be of great motivation and help in springtime!