[FIXED] -> Added dampening rubber mounts (aka rubber bobbins). Glitches disappeared.

Gaak! My new build is acting oddly. When I throttle up beyond a certain throttle-level, the quad starts twitching. You can see in the video below. This happens in Angle mode as well as in Rate mode, however the motors spin up normally when I do so from the motors tab in Betaflight. So... flight controller issue?

I flashed a clean Betaflight 2.9.1 to the flight controller. Since the twitching I've started playing with lowering the PIDs and also applying more low-pass filtering on the gyro and dterm but no changes in the twitching. Ideas??

setup:
frame: GEPRC ZX5"
flight controller: Naze32 Rev.6 / Betaflight 2.9.1
esc: Racerstar v2 30A (BLHeli_s)
motors: DYS SE2205 2300kV
PDB: Realacc HubOSD (pdb+osd)
vcam: Runcam Swift
vtx: Eachine ET200 200mW
battery: 3S Multistar 1400mAh

Check motor direction, then undo all the adjustments you have done - PIDs and filtering, ie put back the defaults. Put some propellers on it (again check direction) and try to hover it in a safe open area. My guess will be that it will fly just fine (for a default tune)

You cannot judge the vibrations caused by running PID stabilization without propellers - that is reason enough for the quad to twitch. This is like revving a car without the tires and asking why it is skidding around. The PID algorithm is freaking out because it needs propellers to manifest the thrust it is actively calculating to achieve stability. When that thrust is not manifesting in the gyro, it thinks the motors need more power, which at some point will just make the stator twitch back - third law of physics

If you need to test the motors individually - use the motor tab. Beyond that, without propellers you can only reasonable test if the quad arms and disarms, and of-course failsafe, which should be tested on every new build before flying - arm it, shut down your radio and see that the quads disarms relatively quickly (1-2 second).

Hope that helps. In the odd chance it does not hover, perhaps the gyro is sensitive or broken, but you can never jump to that conclusion from running it without propellers on a desk

OK, fixed it. Really the *only* flightcontroller-related thing I hadn't brought along from the previous build was the rubber spacers I had been using, which are sometimes called rubber bobbins. I've been using them on my builds for almost as long as I've been in the hobby (6 months), just cause I had read about them somewhere and thought "well, why not ey?", and had never realized just how effective they are regarding the Naze32 Rev6 (mpu6500)!
On this particular frame I had just wanted to follow the frame's build as closely as possible so I left them out. But MAN... the difference. Check out my two blackbox logs of the before and after: https://www.dropbox.com/sh/8ewkgknrbityr...ppTia?dl=0.

So yeah.. rubber bobbins for damn president when it comes to sensitive flight controllers. Get yours here: http://www.banggood.com/4pcs-M3-8x8mm-Ru...63224.html

Here they are, mounted between the PDB and FC.



Jordan, thank you for the insights regardless of my fix. I guess you could be right about the necessity to add props in order to judge the situation, yet I've never had these glitches before without props. Your thought process on how the pid loop could start freaking out because the quad isn't moving towards its intended level would only be the case in Angle mode, not Rate mode, correct?

Hi bombarie, glad you sorted it out. I use those rubber mounts too on many of my quads, they help even with less noisy gyros and keep the motors cooler.

The PID loop runs in Rate mode as well, otherwise the quad cannot fly. It reads the gyro and makes adjustments to the motors as many times per second as you have your looptime set In Angle mode it runs in combination with a self-leveling mechanism that also monitors the accelerometer sensor. In either case, vibrations from the desk or from the motors can start feeding back into the gyro causing the quad to try to compensate, which it cannot without propellers.

However testing the motors using the motor tab and slides does not engage the PID loop or any of the sensors, it is the best way to examine motor/esc only issues. As always keep the propellers off when not flying.