Hexacopter motors load imbalance

hello!

We have a hexacopter running pixhawk 2.1 with DJI ESCs and motors (believe E300)

I have checked the balance along the three motor axis and it is very well balanced.

Otherwise it is quite symmetric, so I hope moments of inertia are balanced too.

What we see is that three motors (alternating, i.e. ones spinning in the same direction) are sent higher PWM than the other 3, consistently.

Here is an example log: https://logs.px4.io/plot_app?log=9eca6934-b657-4976-a32f-b2e56535f05f
It was for more or less stationary flight, just hovering in one place.
Note in ‘Actuator Outputs’ three lines separate from the other three.

Anyone seen this before?
Any ideas why this may be happening consistently?

PS: we are on 1.8.0

Calibrate your ESCs with a parallel (6-way clone) Servo leads using an RC radio … there was a post on this in the past, IIRC the esc_calib code had to change the timeout for DJI motors … saw this in earlier firmware with t-motor ESCs … anyhow, sounds like ESC calibration issue …

Hi, notice in the Actuators 0 plot how the yaw control signal is offset
https://logs.px4.io/plot_app?log=9eca6934-b657-4976-a32f-b2e56535f05f#Nav-Actuator-Controls-0

That could indicate imbalance in yaw, e.g. sometimes people tilt the motors to get some more yaw authority. If not all motors are tilted correctly or by the same amount you can get such imbalances. It could of course also be related to ESC calibration or differences in propeller performance.
I would start with the ESC calibration as a first step.

@jfd - we did run ESC calibration and we are on the latest firmware, so I do not think this is the issue.

@tumbili - I think that’s it - some rotors tilt! We started with standard DJI F550 frame, but then replaced arms which we took from another similar frame. But the screws did not go in perfectly on some arms. So I suspect some arms have 1-2 degree tilt and that would explain the yaw generated.

Ah yes, makes sense. Thanks guys