I’m struggling with AutoTune.
I’ve done it w/ 3.3 a bunch of times, even on this same copter.
W/ 3.4, I have trouble getting it to start… Has the throttle deadband decreased, so it has to be exactly center to run? Or do any other conditions need to be met before it starts?
And I just had a flight and wasn’t sure if AutoTune ran at all… But when I reviewed my log, it looks like for about 5 minutes it didn’t do anything, then finally (not sure what changed) it started (went from 0 to 1) and 8 minutes after that, it finished… I had landed the copter and re-taken off in that time, and that AT status stayed at 1, and finally finished and went to 2.
But, during that time, I noticed a few very small twitches, but nothing like 3.3 with its regular back and forth, increasing in amplitude. I only noticed a few small twitches the entire time it ran.
And because I didn’t know it was running or had completed, I didn’t save my parameters…
Would it be possible to get a tune status display in MP? It would really help a lot!
Cala, it does seem like we’ve got the same problem… If it is related to our ‘racer’ style copters, then it is still something new w/ 3.4. I can auto-tune this same copter with 3.3 and it works perfectly.
Thorsten’s trouble might be related, but I don’t have any complete tunes, so don’t know if they work or not…
Iseries, there’s no way I could hear the buzzer over the sound from the propellers. (These small copters are actually louder than big ones)
Yes, my latest logs with some manual tune are the Diy drones ones, I nade some changes that Leonard recommend looks that He ask for bistmak changes, shure that way there is more info in logs to help developers find the issue, if you can try to do it too, more info looks better to find a solution.
I noticed with small copters that you have to make sure your thr_mid and thr_min are set properly - if you don’t have a good range between these two then the autotune twitches are tiny. I think they’re also smaller with oneshot125.
I have oneshot 125,I’m going to check thr again, another thing, I can’t calibrate esc by MP, I used the old method but motors look’s spinning well and copter flies reazonable. Someone else with oneshot 125 have troubles with autotune?
My thr_mid and thr_min seem to be set properly…
about 300 for mid, and 50 for min.
Alt_Hold is no where near as stable as 3.3, but it does work… (with the transmitter centered, it seems to go up and down a meter or two for a minute, and then finally stabilizes and hovers as it should)
I do have oneshot enabled.
I also can’t calibrate esc’s w/ 3.4.
I plugged them into my auavx2 with 3.3 and calibrated them there, and they seem to be flying fine w/ that calibration and oneshot enabled.
All the suggestions seem to be about the copter or our settings, but I have another one of these same copters, same frame, same ESC’s, motors, everything the same, all settings and PID’s started off the same… But it’s an auav-x2 and copter 3.3. It works perfectly.
The one I’m having trouble with is a pixracer… I’m not ready to put 3.4 on my working auav-x2 copter, although that would be a great test…
My thr_Mid was not set correctly. It was too low in the log above. (I put different props on this copter last week and didn’t adjust it.)
I fixed that, and have a little less trouble starting the tune, but the rest is still the same.
It doesn’t move much, so I don’t really know while in the field, if the tune is working.
Looking at my log, it looks like it did start, but I flipped the switch off after a while because I didn’t think it was doing anything.
What about mot_spin_armed? i had the same problem but was changing all sorts of options to fix oscillation, and one of them also made the autotune twitches much more noticeable.
What do you have AUTOTUNE_AGGR set to? Try changing that, it alters the size of the bounceback. Also what are our current roll/pitch PIDs? If they’re too small then you might not see much movement - does your copter react responsively to stick movement?
Hi, thanks for the suggestions.
Mot_spin_armed is at 55
Autotune_Aggr to .05
Current PID’s are low, because I copied settings from my working 3.3 copter, and manually tuned a little, first, to get the bad oscillations out b4 the auto-tune.
Roll
P,I: .035
D: .0015
Filt 80
Pitch
P,I: .09
D: .003
Filt 40
Copter fly’s great w/ these PID’s… I’m just testing everything and reporting my results… So… Don’t fix this for me… Fix it for the community
What size is your copter, fnoop? (The one you’ve tuned w/ 3.4) Also a little racer size one?
Try changing the autotune_aggr to 0.1 and see if it makes any difference? If you’ve got it flying great, then great! I think up until recently apm just wasn’t used much on smaller copters so most of the defaults/code is geared up for larger copters. Perhaps autotune just needs a bit of tweaking for smaller craft. I’m flying a 360, detailed my issues here: http://discuss.ardupilot.org/t/small-360mm-quad-oscillation-problems/8365
Flies great now
BUT… (this is the part that’s bothering me)
This same copter auto-tunes great w/ 3.3… so they did have it working with these small copters.
Otherwise I do agree with what you’re saying… I think most of the development is geared towards larger craft, which is why we’re finding the issues on these small ones.
But they’re apparently trying to add more support for small copters, so I’d expect the newer release to be easier to set up… not more confusing. (so I’m still hoping it’s something in the code they can change… if not, and things just have to be near perfect for these small copters now, ok… but lets have an official announcement about that!)
–This isn’t worth its own post, but if someone knows the default imax settings for 3.4, could you let me know?
Thanks you!
no, I guess it’s a little misleading to say it’s the “same” copter. Everything is the same other than the flight controller. The copter that works is an AUAV-X2 w/ 3.3.
Autotune does not work on a DJI F550 Flywheel with 1.9 kg.either.(No twitches to see)
And I had to increase the PID’s for Roll/Pitch/Yaw 4 times higher than the defaults to control the copter as it
was with Pixhawk and 3.3.3 This is the result of 30 testflights with about 1 1/2 hour.
The reaction on PID corrections are realy strange and no typicaly. e.g. Start wobbeling at Rate Roll/Pitch P = 0.7
(default Pixracer 0.135 and with Pixhawk 3.3.3 P=0.18, same Frame)
So far no issues with Pixhawk and PX4Lite gold with FW from 2.9 to 3.3.3
