Crazyflie stability

[Capri-Sjonnie]

Hi!

I want to compliment you guys on the fantastic job of creating this! I was really suprised by the overall stability of the flie.
Although I have an issue with the stability when I move quickly back-and-forth or let the flie drop a couple of feet. It will start to wiggle and will level itself out eventually, but still not something I want (control is notably less ofcourse).

Is this a known bug, or is it something you guys have encountered before?

Thanks!

[Surrender]

This issue is also with my crazyflie, most of the time it was not able to level back and it crashed on the floor, struggling and no chance to control it.
It looks like a oscillation where the PID is not able to get out.

Happens when i drop the crazyflie and then give full or nearly full thrust to move up to the old position.

[marcus]

Hi,

Yes, this is something that we have noticed but haven't had the time to investigate yet. We aren't sure how stable the Crazyflie is able to be under these circumstances. I've added an issue to investigate it.

/Marcus

[tobias]

I'm pretty certain this is because the Crazyflie sometimes is outside the stability zone, the motors are then unable to produce the thrusts that are necessary to stabilize it. One reason is because the output of one or more motors is reaching max or zero which can happen e.g. when the Crazyflie descent quickly. The trick is to try and keep it as much as possible within the stable zone and this I'm pretty sure can be improved.

[omwdunkley]

I have no idea if this applies to tiny quadcopters, but descending into ones own downwash is a big deal for helicopters, usually resulting in the rotors generating very little lift. And as lift at each rotor is the only thing keeping the CF stable, that might have something to do with it? Just a thought

[Fucitol]

Also experiencing the same stability issues. Because I'm a new pilot I left the max thrust at 80% and set the max roll/pitch to 15 degrees / sec and added a 3D printed projection frame which lowers its inertia further. Even with these settings, if I switch axes for Roll I often experience osciliations and having it drop altitude rapidly. With my thrust setting I can hardly keep it in the air but if I do it does stabilize again.

[Surrender]

I've found some good information about this in the Arducopter wiki:
APM2_Quad_Stabilize_1.pdf
Tuning

This wobble effect can be fixed by changing/adjusting the PID settings.
According to the PDF linked above we have to increase the P setting:

b. If your Quad wobbles when descending quickly in Stabilize mode increase the “Stabilize Angular Rate P” setting above the default “0.18” (Try “0.20”).

According to the Website "Tuning":

My copter gets wobbly when descending quickly: Raise your gain in Rate Control P. You can tune most of this out, but it’s impossible to descend into your prop wash without some wobbles.

We have to test it out (also because we have other values).
More or less all crazyflies are the same so if these values (finding with trial & error method) are working well then the others can add it too.

Tuning Guide: DIY Drones Tuning Guide