CTP Autopilot problems

[Bill O]

Originally Posted by a64pilot

Curious, why would you think the CPT controller would be any different?

Probably more than this but will take a stab at why the pypilot would be better/more efficient

1. more efficient motor controller
2. more advanced/efficient chips and board parts
3. better programming/software
4. better gain controls for increased efficiency; able to dial in better for conditions/individual boats

I'm sure there are more but those immediately came to mind.


Bill O.

[MartinR]

Originally Posted by a64pilot

Curious, why would you think the CPT controller would be any different?

The CPT has only one gain setting, probably for the proportional gain. And dead-band of course. This often ends up having the controller giving several impulses to the motor for a course correction.

The pypilot gain settings are much more tweakable. There are settings for various gains.

https://pypilot.org/wiki/doku.php?id=tuning_gains

There is a downside to this, though. Incorrectly adjusted, it can drive the rudder excessively. and to use it efficiently, takes a little bit of experimenting, while the CPT with gain set in the middle will work nicely just straight out of the box.

[a64pilot]

See I think the control electronics are insignificant in power consumption, they consume so little it’s just not relevant.
What are you referring to as gain? Speed of the motor? The CPT has two adjustments. The dead band or sensitivity if you will, and how big the correction is, which I’d call gain., but it calls rudder.
It works in pulses and I assume knows rate of turn, and may tailor how much rudder is applied based on how big a direction change is required, or it seems to, but it may not.
It is a simple system

[MartinR]

Originally Posted by a64pilot

See I think the control electronics are insignificant in power consumption, they consume so little it’s just not relevant.
What are you referring to as gain? Speed of the motor? The CPT has two adjustments. The dead band or sensitivity if you will, and how big the correction is, which I’d call gain., but it calls rudder.
It works in pulses and I assume knows rate of turn, and may tailor how much rudder is applied based on how big a direction change is required, or it seems to, but it may not.
It is a simple system

The CPT (modern version) has the same sensor suite as the pypilot, gyro and magnetic compass. So it uses rate of turn and deviation from course to adjust the rudder.
But you really have only two adjustments you can make.

In my opinion, the CPT controller is a good one, but the pypilot is more adjustable.

Also, the pypilot software can be updated... . Hopefully, we will get adaptivity, like in commercial autopilots, where the settings are adjusted automatically to the conditions.

[Bill O]

Originally Posted by MartinR

Also, the pypilot software can be updated... . Hopefully, we will get adaptivity, like in commercial autopilots, where the settings are adjusted automatically to the conditions.

Sean continues to improve his software/features based on feedback. He is also working on a true "learning" computer which will adapt to conditions and the boat.

At this time, I don't know of any commercial system that really automatically adjusts settings and learns. They give you that impression, but you still need to manually adjust gains, etc. for different conditions. Too many variables that aren't constant and a 9 axis IMU can only do so much.


Bill O.

[Exile]

Originally Posted by Bill O

Sean continues to improve his software/features based on feedback. He is also working on a true "learning" computer which will adapt to conditions and the boat.

At this time, I don't know of any commercial system that really automatically adjusts settings and learns. They give you that impression, but you still need to manually adjust gains, etc. for different conditions. Too many variables that aren't constant and a 9 axis IMU can only do so much.


Bill O.

At least some of the brand name APs for the leisure market are advertising "adaptive" systems, but that may just be advertising.

[Bill O]

Originally Posted by Exile

At least some of the brand name APs for the leisure market are advertising "adaptive" systems, but that may just be advertising.

Haven't looked at them in a few years (I read the various usage manuals), but that's exactly what they were saying then. I'm doubting anything has changed since then. At this point, none are truly learning and would need more sensors to be adaptive to complex wave patterns, etc.

Heck, most of the big brand name APs are made for powerboats and then they threw in some extras for sailboats. At least pypilot was created by a sailor to be used primarily on sailboats.


Bill O.

[Exile]

Originally Posted by Bill O

Haven't looked at them in a few years (I read the various usage manuals), but that's exactly what they were saying then. I'm doubting anything has changed since then. At this point, none are truly learning and would need more sensors to be adaptive to complex wave patterns, etc.

Heck, most of the big brand name APs are made for powerboats and then they threw in some extras for sailboats. At least pypilot was created by a sailor to be used primarily on sailboats.


Bill O.

I've read about compasses going from 3-axis fluxgates to 9-axis gyros but I haven't looked into what this really means. My old B&G below decks Hydra Pilot has always impressed me, until it started getting glitchy that is. Probably just age.

[a64pilot]

I know my compass if it were in an aircraft would be called an AHARS, or advanced heading attitude reference system.
Mine knows heading, pitch roll, and I assume yaw, it may not know yaw without a ground track, but it does know pitch, heading and roll. I believe it also knows accelerations.
This guy
http://www.airmar.com/productdescription.html?id=94

Geez my anemometer puts temp, wind speed and direction and barometric pressure and relative humidity on the bus, you can therefore compute density altitude for a true airspeed if you wanted to.

I’m guessing these things do all of that because the chips are available and why not?

[Bill O]

Originally Posted by Exile

I've read about compasses going from 3-axis fluxgates to 9-axis gyros but I haven't looked into what this really means. My old B&G below decks Hydra Pilot has always impressed me, until it started getting glitchy that is. Probably just age.

The 9-axis inertial motion sensor (IMU) includes a 3-axis gyroscope, a 3-axis accelerometer and a 3-axis magnetometer is in a chip less than the size of a quarter.
My IMU chip contains a barometer also.

[hpeer]

I’m still in the early phases of using a PYPILOT. Having a steel boat my compass makes a great coat rack. Making the run from Grenada to St Martin I saw the binnical compass do nearly 180° swings on more than one occasion. However the PyPilot compass, mounted in a wooden partition below decks seemed to work fairly accurately. The (old style) CPT control head could not lock into a usable magnetic heading and would set the boat to spinning in circles, constant left rudder.

Going South things were better and I could use the CPT but course changes were not proportional to the actual change.

This may well have been resolved with the new CPT control head.

The Pelagic control head seems harder to calibrate than the PyPilot, that requires you to do a 360° durn marking 15° increments. I find that difficult to do, especially without a compass. The pelagic is on our small boat which is also steel. It steered just fine but Even a small 2° correction would move th e heading 10°. I talked to Pelagic and they suggest I do the calibration over again.

[Exile]

Originally Posted by Bill O

The 9-axis inertial motion sensor (IMU) includes a 3-axis gyroscope, a 3-axis accelerometer and a 3-axis magnetometer is in a chip less than the size of a quarter.
My IMU chip contains a barometer also.

Helps explain why Sean was able to make the PyPilot components so small.

[a64pilot]

What is honest to god scary is when you learn the accuracy of these small, inexpensive chips, it’s not been that long ago that to get almost as good as these quarter sized chips required a box full of accelerometers and gyro scopes that took six to nine minutes to align, was about a foot and a half square and cost as much as a nice car, and you couldn’t move it during that time, and even then it needed to be slaved to a Doppler Radar or it would go into free inertia and errors would begin to compound.
1980’s AH-64A, the 1990’s AH-64D had dual ring laser INU’s that erected instantly and used GPS, and I’m sure is solid state.

I’ve never seen one, but I believe the early submarine inertial Nav units were room sized.

[MartinR]

[QUOTE=Bill O;3030551
At this time, I don't know of any commercial system that really automatically adjusts settings and learns. [/QUOTE]

Sorry, I was not clear. With commercial I mean for commercial vessels. Started in the 1980s with the Kockums Steermaster and now many autopilots are adaptive. No gain settings anymore.

I can see no reason why this technology could not be transferred to leisure craft electronics. It is mostly a question of software.

[hpeer]

Originally Posted by a64pilot

What is honest to god scary is when you learn the accuracy of these small, inexpensive chips, it’s not been that long ago that to get almost as good as these quarter sized chips required a box full of accelerometers and gyro scopes that took six to nine minutes to align, was about a foot and a half square and cost as much as a nice car, and you couldn’t move it during that time, and even then it needed to be slaved to a Doppler Radar or it would go into free inertia and errors would begin to compound.
1980’s AH-64A, the 1990’s AH-64D had dual ring laser INU’s that erected instantly and used GPS, and I’m sure is solid state.

I’ve never seen one, but I believe the early submarine inertial Nav units were room sized.

Mid to early 1970’s I flew as radioman in the flight deck of USCG C-130’s. We had LORAN A and C and a few planes had inertial nav. It took at least 15 minutes of sitting on the tarmac With the thing running to get the system to stabilize. And they could very infrequently develop a cumulative error.

So I’m very much in agreement with you comment about the miniaturization of both size and cost. Truly amazing.