Need to see a log with full rate attitude logging. But but it sounds like noise…
You are running with the harmonic notch presumably?
Thanks Andy. Yes, I am running with harmonic notch. Still waiting on SD cards to be delivered for logging ability!
I’ll change Acro_Option to rate only (per Leonard’s recommendation above) in the meantime and see if that helps. It’s true what he said, I don’t have the tune locked in yet. Did Autotune and needed to back off PIDs results. I noticed this strange behavior when actually working on the tune.
Thank you Leonard. That may be what I experiencing. I do not have tune yet where I want it to be, still working on it. I did notice this behavior when working on it. I’ll change Acto_Option to rate only and see what happens with that. Thanks again for your help!
What you describe sounds like the difference between a good tune and a bad tune
Yes, the “tune” I have in it now is not ideal as I’ve stated. Still working on it (blindly without logging capability yet, no SD card).
The “uncommanded movement” I have experienced I was not expecting. I’ve had less than ideal tunes before on other firmware and have never seen this.
I’ll change AcroOption first to see if that has any effect. If it does, then that would be a contributing factor. If not, then the slightly modified Autotune I currently have on it is much worse than I thought (flies nice apart from this one type of aggressive maneuver). Thanks again Andy.
If everything is perfect sure. The integral of velocity is position. So the aircraft will need the same total rotation to unwind the I term as it did to create it… unless:
- you have more than one axis doing corrections at the same time,
- you don’t hit I_MAX
- there are no gyro bias
- there is no aliasing from higher frequencies
If everything is perfect sure. The integral of velocity is position. So the aircraft will need the same total rotation to unwind the I term as it did to create it… unless:
- you have more than one axis doing corrections at the same time,
- you don’t hit I_MAX
- there are no gyro bias
- there is no aliasing from higher frequencies
Right, and since it’s unlikely everything lines up perfectly I would expect the attitude(angle) will drift from external forces if we’re running rate-only mode. Which makes that rate-only mode not all that useful for normal flying?
Beta flight run rate only mode. So if you use a high enough I term then Rate only mode is very useable.
Yes sir! I’ll try this out 
Just keep in mind that our standard tuning instructions are for an attitude stabilized system so we run very small I terms because the Attitude P term does most of that work. The two are a little different when we do dynamic manurers in how they respond.
ATC_THR_MIX_MAN? Options are 0.1 to 0.9. Do you mean 0.2 - 0.4?
Thanks John
You can go greater than 1… Up to 4 I think.
Mission planner says range is below 0.9 but that’s just a warning. The reason to keep it below one is so oscillations wouldn’t make the quad fly away (ie.motor authority is above hover point) but if you don’t have such issues it’s fine to go above 1. Also it causes bouncing when you land. So usually I land in stabilization mode where mix_man isn’t in effect.
Alright Andy / @Leonardthall , I’ve got logging capability now and have included a test flight log here if you’re willing to take a look at it, I’d appreciate it. It’s Acro Mode.
Still having those “uncontrolled movements” on aggressive pitched-forward turns.
As @chronowarper thought, it was not fixed by Acro + Rate Loop Only. I now understand his “flew like a drunken pilot” statement!
So, I believe you are correct in that I (at least) have a poor tune currently.
You can see the Roll P term with extreme overshoots on the charts, however, I’ve reduced this term from the Autotune value of 0.08358493 to 0.07. It also looks like these overshoots are happening primarily only in one direction (roll right)?
Also, still too much P (or not enough D) on Pitch axis?
I’ve also attached the Autotuned Roll and Pitch PID values image here and my current PIDs used to create the test flight log here
As always, I appreciate your time and thank you for any assistance you’re able to provide.
Ahh, I see. I thought those were hard limits.
I’ll give 3 a shot
Alright John, set it 3 and gave her another shot.
Guess what? “Uncommanded movements” seem to be gone (or at least almost gone, might give a value of “4” a try to see what happens)! I wouldn’t have thought of this without your help, thank you so much!
I still do have tuning issues (which, by the way, you’re more than welcome to view those linked files I posted earlier for Andy / Leonard) if you’d like and offer suggestions.
If you don’t mind, how much throttle offset did you add for Acro and is it only in affect for Acro or did you just add it on the Throttle channel lie in Inputs (i.e. affects all modes)?
I just now started reading through your entire thread here. Watched your short video clip from October 4th on zero throttle authority loss. “That huge wobble at 00:05” is pretty much the same thing I have been experiencing except more dramatically/quickly in a more aggressive rotation with nose down and throttling down for a quicker 180-type turn. As I just posted a few minutes ago, your ATC_THR_MIN_MAX of 3 fixed it for me. I’m gonna look at throttle offset also
Your basic problem is that you are not using the harmonic notch (surprise!):
- You have INS_HNTCH_FREQ = 200, this means the notch will never go below 200Hz, unlikely that your copter has such high kv motors that this is reasonable. Set it to something like 60Hz.
- You have no ESC data so the notch is not moving at all. If you are expecting bdshot to give you this then you need to set SERVO_BLH_BDMASK, if you are expecting ESC telemetry to give you this then you don’t have it set up correctly.
2a. If you are trying to use ESC telemetry then SERVO_BLH_TRATE is too low - you need something like 100 (but leave as is if using bdshot). - You are using the double-notch - unlikely that this is helping other than to introduce latency. Set INS_HNTCH_OPTS=2.
1 Like
Well now, that harmonic notch is working better! 
I enabled the SERVO_BLH_BDMASK . Didn’t see that anywhere in the setup for the harmonic notch on the Ardupilot.org site.
I’m running 1504.5 2950Kv motors
I’ll set INS_HNTCH_OPTS = 2 as you suggested also. I noticed the delay as well.
Thanks again Andy!
Take care