last week I tried autotune with my Y6 copter. With my tricopter it worked so great, that I was expecting a good result with the Y6.
At some point the copter did some strange movements and nearly crashed to the ground. I could recover it by taking over controll and did an emergency landing in a field.
Ii was very foggy that day and after the flight, the copter was pretty wet. So I thought, it could be some issue caused by the wetness.
So I did not pay to much attention to that.
Today I took out my Y6 and tried autotune again as it was dry and no wind today.
I started autotune and it began to do the rocking over roll and then over pitch. Nearly at the end of the pitch session, it became really scary. The back motors nearly stopped for several times what led to very scary movements. I had enough height and good nerves and so I let it happen. It was exactly the same as it was last week when I did autotun in foggy conditoins.
After a few of this movements, it stopped. Then I flew a few rounds, just to give the compass same data to calibrate and then landed, disarmed and switch off autotune to save the values.
Then tried the new settings and was really disappointed. the copter was terribly shaky.
At home I looked at the PID values and saw some really strange values.
I attached the PID settings before the autotune (Defaults with 3.1rc5 Y6) and after the autotune.
I also attached the logfile of that flight.
I mam just uploading the onboard video. Will link to it, when it's all up.
Hopefully someone could find out what's going wrong with this autotune.
Here is the onboard video.
I just red a statement from Leonard Hall where he says (besides some other things): "Thanks for all the people giving auto tune an early run! And please get hold of me over at DIY Drones with any problems you find."
Well, I am not an english native speaker, but sounds to me, that it they would like to have some feedback.
So I posted my feedback including logfile and video, but nobody answers.
For my Y6 I decided to let him with the default PIDs as he flies very well with that, but I'd like to know what is going wrong with autotuning on that frame. It worked really well with my tricopter, so I was really excited about that.
Next week I receive a new flat hexacopter frame with new motors, ESCs and props. Hopefully autotune will work fine on that frame.
But until then I really want to understand what happened during that autotune session.
I don't see new discussion posts but I do reply to any autotune questions on the latest release thread. That is how I found your post from 9 hours ago (about when I went to bed last night). I am also a single father with 2 children under 3 years old and my hobby has to come second. So you will have to forgive me for the delay in answering your question.
You are correct that Autotune has not worked for on your frame. I have seen this problem before and I believe it is caused by an overly flexible frame or overly soft vibration isolation on your APM. I would suggest that you see if you can move the arms or any of the center plates and check that the APM feels solid in the way it is mounted.
Other issues that can mess up Autotune is an incorrect CG, props being too large for the motor and battery combination, and letting the copter build up too much airspeed during the test.
If you post some photos of your frame here I may be able to make other suggestions.
Now that I have replied to your post I should get an email when you reply.
first of all let me say, that you and the other guys from the developer team have my full respect for your work.
I understand, that it is not possible to read all posts here and answer within minutes.
I just was a little bit astonished, that a detailed error description of a new feature did not catch someones interest.
Now I see, that the Arducopter 3.x part of the forum is not really watched. It is a little bit strange that all the 3.x related issues are discussed in the 2.x section.
Now on topic.
My frrame is a self built frame with wooden centerplates and 10x10mm aluminium arms.
It's weight is exactly 1.8kg including 2 pcs. 2200mAh 3S batteries and including a 198g videocamera.
The CG has to be o.k. and the props are 10x4.5. So they should fit.
I attached some photos to this post.
Maybe I didn't make myself clear. You said, "Now I see, that the Arducopter 3.x part of the forum is not really watched. It is a little bit strange that all the 3.x related issues are discussed in the 2.x section."
I saw your post on the 3.0.1 release thread and that is how I found this thread. I don't watch the 2.x section anymore. The general approach is to post quick questions directly on the 3.0.1 release thread or do as you have done here and post a detailed description on a separate thread and get our attention to it on the 3.0.1 thread. What you have done is my prefered approach. So thanks for making the extra effort!
So looking at your frame I think I might know what is happening. That lovely camera mount with all the weight on it is not moving with the frame and is causing the wobble I am seeing in the tuning logs. I would suggest you do your autotune with your batteries hard mounted to the frame, without the camera, two carbon tubes and rubber mounts (or with the batteries as hard as you can make it on the carbon tubes). This will give you a tune in the most responsive condition of the frame that should then be a very safe tune for the copter.
If you do try this could you report your results back here. This is a very interesting data point for me because many people will be trying to tune with similar setups.
many thanks for your answer. That sounds logically to me.
I thought it would be a good idea, to have the same setup when tuning, as I have when flying.
Next time I will hard mount the camera (or is it better to to it without camera?) and the lipos to the frame and will give that a try.
Camera will be turned on and I wil luse an additional groundcamera.
Are the logs configured o.k., so that you can see all what you want tosee or should I enable something else or disable something to avoid performance issues?
A question to the values after autotune. Do you think the stab PIDs cause the problem or is it the rate PIDs?
Maybe I could use the PIDs from autotune and just reduce stab PIDs to the defaults.
Yesterday I flew the Y6 with the defaults as they are in the Y6 code and it flew pretty well. Only some wobbling when descending, that has some space for improvement.
Just another reply regarding the 3.0.1 thread. Yes, you are right, I posted my post that brought you here in the 3.0.1 thread, but this thread is in the multicopter2.x section of the forum.
I find it a little bit confusing that the 3.0.1 thread is in the 2.x section. Hopefully the 3.1 thread will be in the multicopter3.x section of the forum.
You probably have the answer to this already, but you can search for "autotune esc sync issues" to find a complete explanation of the issue. Also the Marco Robustini sync problem video at https://www.youtube.com/watch?v=0lRTN5o4UCM
Thanks for offering help here but this isn't the esc sync issue. This is the large flexible mass issue with autotune.
OK thanks Leonard, I saw his initial comments about some motors nearly stopping. But yes there looks like a lot of compliance between the airframe and the extended mass of LiPo/Camera based on the photos/video. Thanks again. Richard