mpu broken 3 times?

[sideluxe]

Hi threre, 

after a long time without getting the project forward i wanted to update to the newest build and software.
Last year the pypilot started working slowly but had some space for improvement on my boat.

2 things i dont understand actually are

1.  after updating via git everything seems to work but after a reboot my display hangs on pypiolt 0.12 and the keypad is not working using the autopilot via web control or plugin is still possble

2.  yesterday all my mpus stopped working, heading is still displayed, but when turning the heading changes to a new direction and then goes back to the old heading before turning.
i tried 3 different mpu as well as pypilot 0.9, 0.12, 0.15 all the same
python boatimu.py has gyro and heading values but heading changes as above when tunring the mpu.

i dont know where to search next, because i dont think all mpus are dead. could this be a defective pi zero too?

Any suggestions would be fine.

Thank you all

[seandepagnier]

it sounds like the first issue with update doesn't actually update the splash screen.

As for the compass... I recommend running pypilot_calibration <ip> on a linux host so you can visualize the compass readings and ensure the calibration is correct and the boat is level.

[sideluxe]

Hi Sean,

thanks for helping out.
The imu problem seems to be fixed.
It works like it should now, perhabs the cal was totally faulty.

Now I have another problem again, enabling the ap first moves the motor about 2cm in one direction, which is bad and staying on, the motor moves backwards periodically without changing course or doing anything. Depending on d gain more fast til it reaches port fault.
Any ideas on where to search?
Compass readings seem stable, changing heading or desired course also seems to work. But the rudder does not stay on course in any circumstance.

Further, do you have any idea why the git update crashes lcd and button function in my case.

If I could post any logs or so, please advise.

Best and keep up the good work

Christian

Sent from my BLA-L29 using Tapatalk

[seandepagnier]

I'm not sure why enabling the ap causes movement, but it could be several reasons.

Can you look at the gyro output? It takes a while for the bias filter to compensate so that they reach zero.

I'm not sure why it might crash but there have been a lot of recent changes in git since the last stable image.

[sideluxe]

Hi Sean,

I got the time to dig a little and got my display working again.
Looking at the service logs it was obvious that using a git clone and pypilot.build will  delete the needed folders lcd, rf and webapp within the site packages dir.
After copying everything back in place the logs are clear and everything works as expected again.
Got no time to take a deeper look at the script and what causes it.

JUST SAW THE LATEST CHANGES AND COMMENTS ABOUT NAME AND DIR CHANGES...
So the services side of tcz is not changed correctly then, i assume


About the short movement I am note really sure war causes this by now. I updated all to latest Git including motor.ino and now it only  shows this behavior when enabling the ap for the first time, afterwards everything looks as normal it could look at a desk test.

Perhabs you could answer a short one about the imu, after calibrating the accels it shows one 0.000 and one 1.010 value.

Is this acceptable..?
Alle 3 imu show 0.000 on the last axis after cal. The 1.010 is only on one.


Cheers

Christian


Sent from my BLA-L29 using Tapatalk

[seandepagnier]

Yes, some services changed names since the last image. The next image will update the services.

As for accel calibration, yes it's normal to have values near zero for the first 3, and the last is near 1.