Random undesired hard roll

Hello everybody, we have 14 identical 700mm quadcopters, hardware is the same, configuration is the same, most of them fly pretty well, but some of them aren’t stable at all.
I am attaching a log of the flight and a video Drive folder, containing a video of the flight and the log of the same flight.
I will be very grateful for any ideas what might be causing the problem.

Tuning is is rather poor (hasn’t been done, a lot of defaults) and the Motor Outputs are oscillating.

My guess is that all of them, if they have the same parameters, are borderline poor and some cross over into bad. So, start the tuning process!
Drop the Rate Pitch/Roll P&I values to 0.10 and the D terms to .002
Set these:
INS_ACCEL_FILTER,10
PSC_ACCZ_I,0.4
PSC_ACCZ_P,0.2
INS_LOG_BAT_MASK,1
INS_LOG_BAT_OPT,4

Make a short AltHold hover flight to collect data to configure the Notch Filter and then review the log for output oscillation. If it’s gone then run Auto Tune.

Give this a review too. The MOT_SPIN_MIN looks high:
Motor Rangers

2 Likes

Hello, Dave, thanks for the feedback!
I will definitely try tuning it according to your instructions.
Most of the parameters are default, its true, but I have run all pre flight set up, Motor Rangers, initial parameter calculator etc. The motors don’t react bellow 14% throttle. The others with the same parameters actually fly pretty well in practice ( I know its not expected with the default tune).
I will attach a video and log of one flight with the other quadcopter that flies good (at least from what I can see)another quadcopter, with the same configuration, but flies much better (video + log, google drive)
I reviewed the logs from one of the “fine” flying quadcopters, and saw the same oscillations in the PWM outputs to the motors, but the drone is pretty responsive, and doesn’t wobble.
Also, I noticed a misalignment in the desired vs actual roll graph on the problematic one.

OK, but that’s not tuning. That’s pre-tune to get it in the air to collect data to advance the tuning. You are out over your skiis flying these before a complete tune.

This will bite you at some point. Outputs oscillation is a sign of a bad tune and will often heat things up and they can fail as a result.

Right, that’s just simply a bad tune. You cannot expect to fly on defaults. The process isn’t that difficult:
Get it hovering w/o outputs oscillation.
Configure the Notch Filter. This isn’t optional it’s part of standard tuning.
Run Auto Tune. Unless you have experience with manual Tuning then do that.

2 Likes

I will try to achieve level hoover without motor output oscillations today, and start autotune. Thank you for showing me what to look out as a sign for a bad tune. Last week I tried running autotune according to the guide, but it kept returning “failed to level”, then I tried manually tuning it, but lets say that the result of my tune was even worse than the default(at least it didn’t crash :grin:).
Since then I have done accelerometer calibrations on a perfectly level surface, and will try running ESC calibrations again. I hope that this time auto-tune will succeed.
I will be back with the results of today’s flights, accompanied by logs.

Hello, I have done a few test flights in AltHold mode. I’ve entered the parameters you suggested, made a FFT log, and tried setting up a notch filter, the problematic copter flies a lot better, the oscillations in the motor out PWM are lower, but never the less they are still present, that’s why I haven’t tried running autotune. I am attaching the logs before and after the Notch filters. I want to ask, what is the acceptable level of motor PWM output oscillations, and is the copter ready for autotune?
Thank you for your feedback, it helps me a lot in learning the principles of basic tuning and instrument interference!
Drive containing logs before and after Notch filter

Some you did but you didn’t do this “Drop the Rate Pitch/Roll P&I values to 0.10 and the D terms to .002”

And the Notch center frequency looks more like 50Hz so you can change that.
You can try to run Auto Tune as it’s configured now as the oscillation is mostly gone. If you get an error message saying “failure to level, tune manually” then drop the Rate PID’s as I said.

What ESC’s do you have?

1 Like

Oh, I was pretty sure I changed the PIDs, probably I didn’t saved them correctly, my mistake. The ESC are basic 40A Hobbywing, PWM only, I will check the exact model.
Tomorrow I will be flying again and will modify the Notch filter, and run AutoTune. Is there any good guide to manual tunning of the PIDs, whitch you could recommend me, because I want to learn to do it manually.

This doesn’t sound right. Using Motor Test in Mission Planner they don’t start until 14%? Did you calibrate them?

Manual Tuning:
Manual Tuning

1 Like

I will run calibrations again, and see if there is any difference. We have other quadcopters like Tarot 650 and Align M480, and their ESCs react at 5-6%. I am sure that some of the 14 quadcopters had their ESCs calibrated, I will check their MOT_SPIN_MIN

You should use BLHeli_32 ESC’s. No calibration required plus other advantages.

1 Like