This is a discussion re the ArduCopter-3.1 AutoTune feature.  A place for people to report successes and failures, ask questions and get answers.

The AutoTune wiki page is here.

 

Views: 74305

Reply to This

Replies to This Discussion

Hi Ralph,

Yeh you should be fine to go back to 3.0.1 if you intend to drop the Stab_P parameter (say, less than 6). The only change in the 3.1.rcX code is there to allows users to increase Stab_P to high values safely.

Let us know how your flight tests go!

Hi Josh,

Could you post a log of this mate? I would really like to know why this copter doesn't Autotune properly, or why it did this time.

You have a vibration isolated layer there don't you. Could you replace those vibration standoffs with solid ones and redo the tune?

Leonard,

I put the camera back on and the oscillations returned - although not as severe as before.  Especially apparent during hover in loiter mode.  I have not made any changes to autotune results.  Would this be due to added weight or some interaction between the flight controller and the gimbal controller?

Thanks,

Karl

Leonard, that sounds great. Then we could even assign 1 function to Ch7 and 1 to Ch8 and could tune 2 different functions without pluging the APM to the pc or using telemetry.

Hi  Leonard,

First, thanks a bunch for continuing to help me with this guy!  It means a lot :)

I tried to attach the df log that left the roll pretty pathetic, but it seems to have disappeared.  In FPV, it rocked around like it was a sail boat in rough waters.  The pitch is fine, though.  I spent 5 batteries today slowly bringing up the pids for Roll, but it isn't there yet.  It's so frustrating lol..  I'll get the PIDs close again tomorrow and redo autotune so I can get a fresh df log.

I've did attach the last log from my flights today to give an idea of what I'm seeing there, and you'll see it still isn't doing what I want it to do, but its getting closer as long as I fly like a grandma.

Replacing the Anti-vibe bobbins with hard standoffs is harder than it should be (I don't have any at the minute) but I'll see what I can do.  I think I may have some nylon ones around here somewhere.  Question on that though: before I do that, will the recommendation be that I run the frame long term with those hard standoffs, or will they just serve the autotune process leaving me to replace the bobbins back after it completes, with an ideal flying environment?  If the expectation is that I would need to leave them on, then I wont waste my time - UNLESS doing that provides you with some useful data.  Flying the model long term w/o the bobbins isn't an option though, as it is purpose-built with those bobbins to keep the camera nice and smooth.

Also, I'm pretty bad at listening; I've been grabbing from Trunk lately just because :)  The reason I mention that is because if you decide you want to try some custom loads that may get you more data, or help with fine tuning AT, I'm willing to go down that road.  If you have a way to AT Yaw, I'd LOVE to check that out, because nothing I do seems to nail that one down, as it continues to hunt in fast forward flight (also visible in the attached log)

Cheers bud,

Josh

Attachments:

Hi Karl,

I suspect that this is an interaction between the movement of the frame and the camera, because the camera is not rigidly fixed to the frame. Weather the presence of the camera will cause problems may depend on the frequency that the camera wobbles and how quickly the camera stops wobbling.

I suspect that the reason you are seeing the wobbles is because the frequency of the camera mount is the same as the frame control system and this is causing the oscillations to be amplified.

Could you do an experiment for me. Could you reduce your Rate_P by 30% and see if it still oscillates. If it still does oscillate could you also drop your Rate_D by 30%.

What I am trying to do here is move the frequency of the frame and controller from the wobbling frequency of the camera. Hopefully this will stop your oscillation.

If you do do this for me, please be careful. You should be reasonably safe but you never know!

Hi Josh,

Yeh, I can see what you are fighting with.

It would be great if you could do Autotune without the vibrations standoffs. I am hoping that you can do Autotune without the vibration standoffs then put them back and it will fly correctly.

If we find that it has some fast oscillation I would like to test a theory I have that this is caused by the isolated platform being the same resonant frequency as the control system and frame.

Thanks for your help and perseverance mate!!!

A few questions, just for clarity...

1. Would an AT performed in still conditions yield substantially different values from a WINDY AutoTune?

2.If you got values from the still AT and it flew great, and the copter misbehaved a bit in th windy conditions, which pids would be the ones to change?

3. If the AT dies not use GPS functions, would it be ok to perform the AT in a large indoor arena (metal roof, no GPS receprion)

thanx Guys

1. Yes, but actually because the wind could cause the AT process to fail to work well and not give you good values.

2. I'd guess Stab P, but it really depends on what it's doing, logs needed.

3. Yes... but I don't recommend trying this right now, as there have been a few issues flying indoors.  I'm not sure they've all been fixed up.  I think it can be done but you have to be really careful with setup. For example, one guy tried flying indoors with the GPS installed, and Fence turned on.  It then attempted a Fence RTL and things didn't go well.  Whenever flying indoors, it's probably best to just disconnect the GPS entirely.  But this is kind of a wider topic that needs more discussion with Randy and Leonard.

Hi Leonard,

Took a few flights yesterday and my Y6 definitely feels very much so 'locked in', way more than before. Flew really well! I downgraded to 3.0.1, entered the same values I received from autotune other than I lowered Stab rate/pitch from 8.25 to 7.25.

Only thing I can say I am noticing more than before I tuned is a wobble at times on descent. Below is a video,  flown in stabilize, unedited & no filters (sound is on so you can hear the motors for any twitches) - at the end when I come to land, you can see the oscillations and attempted corrections as I lose altitude.

http://www.youtube.com/watch?v=hO6IZk_72Eo

Having autotune for YAW would really be key for me. I hope it is added to the 3.1 release. Let me know your thoughts! 

I was able to Autotune my quad copter, see below..

I did not have time to fly with the new PID values, is started to be cold here.

Small problems encountered when I try to autotune with my dx6i.

The condition for using APM Ch7 or 8 is to have a value bigger that 1800. My 6th channel from dx6i had a maximum of 1500, even increasing the travel to 125.

Then I did switch my Mode channel (5) with Aux(flaps) channel, 6 and I was able to have 1800.

Autotune process was OK, it did start with some audio message from Mission planner and ended with another message wish Success. 

It takes somehow longer 8-9 min for a 2200 Lipo and I did changed 2 batteries until I finished with Success.

I am curious to see how to fly, I need to switch again my channels as original.

Not sure if new PID values below are good or no, until I will fly again.

But even before Autotune, quad was stable, after a new install of rc5  (after reset/erase). Is is my first rc5, after rc2 before.

New value is before Autotune, after comparing

I miss saving my autotune pid's how can I retrieve it? Is there any chance or should I run the autotune again?

Reply to Discussion

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service