0.56 course holding

[kalle]

I calibrated and locked calibration and trail in calm weather in Front of the harbour. I suppose my problem hast something to do with i2c connection. As I have a steel boat I need an extender. What hast been changed in the last version perhaps i2c baudrate?

[oceandrean]

Sean,

What do you recommend if completing a circle does not give a compass calibration? I have this with all versions. With 0.24 it does not seem to matter.

My gearbox is fixed and I am underway again using 0.24. I have pypilot scope installed. What would you like me to monitor?

Also, to be clear, 0.54 can work well for hours but sometimes starts to steer in the wrong direction. The command heading is correct but it continues to actively steer further and further off course (always to port so far).

[seandepagnier]

The command heading or the reported heading? I keep trying to ask this.

I am wondering if the reported heading is also wrong, or if the commanded heading is close to the reported heading when it steers wrong.

It continues to actively steer... how fast or far? How long is it wrong? Does the compass heading give the wrong heading on some courses or does it change by the correct amount so that it is 90 degrees apart facing north south east and west (+- a few degrees is ok)

Did you level the boat before completing the circle? It is essential. Also some IMU has large accelerometer bias, so you must calibrate the accels. Not really common on the newer icm20948, but even then it can improve things. So I recommend you calibrate the accels, then mount the unit to the boat. Then Level the sensors to the boat, then complete a circle.

It really must update the calibration. Can you see on the calibration page the sphere, it should also report possible errors with calibration there.