Hi all, I've just had two very hair-raising flights. Here is a FPV recording of the particularly bad section of the second one. Note the climb rate (middle left), altitude (top right) and artificial horizon all detaching from reality at about 15 seconds:
https://youtu.be/KbJaZYLe8AQ
I'm flying a White Sheep 480mm quad, with pixhawk and high quality ESCs, motors etc. It's previously flown reasonably well; although it's never been perfectly tuned, it was stable and easily controllable. Tonight I added a cheap Banggood gimbal, and everything went wrong. Each time I put on any speed or power, I got massive uncommanded movements in all 3 axes. These eventually died down both times, but each time I struggled to manage to land it.
I have had a look at the logs, and my vibration seems pretty high, especially in the z-axis. My props are balanced, and the pixhawk is mounted in a decent bed of soft memory foam (10mm thick or so). It's
previously flown fine like this, although I have struggled to keep vibration down. However this behaviour terrifies me, and I won't be happy flying it again until I know what's happened.
The logs also show an EKF error, about the time that the quad went wild. I suspect this is related to vibration, but don't fully understand it - my reading implies that in manual mode, this has no impact on flight, only in automatic modes (where it will force a controlled landing). I was in acro throughout.
The M8N GPS/Compass module I wanted was sold out, the one I have doesn't seem to have a compass in it so I don't have it plugged in, and I'm using the built-in one. In acro throughout so the GPS shouldn't matter, although it has a good lock throughout.
Full logs for the flights are here. One short hover, one terrifying flight. Two more short flights, making sure it was sane, then another lunge of idiocy.
Any help diagnosing what is going on, or advice for how to sort it, would be gratefully received!
Replies
Thanks for givinge me an idea of where to look in the logs too.
Michael, it was certainly worse as the speed increased so I think you are right, vibrations seem to be triggering the bad behaviour.
There is no "genuine-testing" - no need , as it's open source + open hardware.
The problem is, that many manufacturers cut corners, skip components, and reduce capacitor values, even use resistors with high tolerances, and B-grade sensors (that really was discarded as defective because they are a bit outside specs.)
Some are actual *clones* made perfectly well.
I can spot some differences if you post good close-up photos of the PCB inside.
Anyway - as you see, one accelerometer reports crazy vibrations in a whole different range. - maybe, that can be calibrated away..
you have several variance warnings, -recalibrate IMU's. (the difference is so insanely huge, that I am not sure a calibration can fix it, is it an actual Pixhawk (fylly made as the open hardware it is) , or some cheap pile of defective components ?)
the data between IMU.Acc* and IMU2.Acc* is so crazy different, it's not supposed to fly.
you can do recalibration, and hope it's within what can be calibrated away..
Vibrations are high. I have found that flying into the wind causes them to be worse and adding a gimbal is added weight which causes the motors to work harder increasing vibrations to the frame.
I have found that a solid bed of foam transfers more vibrations then just using 4 small squares.
I do not fly in Acro so don't know how the vibrations effect it.