[Solved] Pixhawk2, yaws to left without input (3.5.1)

Hi, i have a strange problem with my big octocopter:

The setup:

  • pixhawk2 + heregps (disabled internal compass)
  • arducopter 3.5.1
  • tiger u7 420kv + 18x6,1 props + tiger 80A ESCs
  • 4x10000mah batteries (15C)
  • frsky 8ch transmitter module and receiver
  • Turnigy 9xpro transmitter
  • Attached a Mövi M15 with about 4-5kg

The setup generally flies great, but during some flights, mostly after descending the copter, it starts turning to left and i have to compensate this behaviour with the yaw-stick until it stops and fiies normally again (some times this takes 30-60secs seconds until i can release the yaw stick again to center). This happens in AltHold and Loiter (didn’t try it in stabilize mode).

I have attached 2 log files where this happend multiple times in flight.

An autoanalysis showed up no errors.
Reviewing the log shows up some things

  • desYaw and Yaw are matching (so i dont think its a pid issue)
  • ch3 (yaw) there are some spikes, but i cannot difference them from my counter inputs to keep the copter stable (could my transmitter be the reason?)
  • is that mag x,y values are aligning with the yaw changes

So my question, is this a compass issue or a TX/RX issue or pid issue?
(Or could that even be the yaw motor of the gimbal that turns the copter that way?)

Perhaps someone with more expirience in analysing logfiles can help me to find the problem or give me some hints where i should investigate

Thanks in advance

I had a similar problem, and it turned out to be a problem with the gimbal on the transmitter itself. The gimbal wasn’t centering properly.

At the time, my 9xr pro was new enough that I was able to get a replacement transmitter from hobby king.

I have a similar issue as well with the 9XRPro that one of the channels is off a little bit from time to time and will cause the copter to drift left or right when the gimbal is centered. I have adjusted the center band to be a little wider so this does not happen and check the calibration screen to make sure the output value has not drifted.


thank you for the hints,
i will try replacing the 9xrpro with a new one in the next days and will report if this will solve the problem.
in addition increasing the deadband a little bit is a good idea too.


Check your Att.Yaw against RCIN Yaw. If RCIN spike goes after the Att.Yaw than this is a counter input.

I had this problem with 2 new builds until I ran AutoTune.

In my case one aircraft yawed left and the other yawed right

The issue was the left stick was not centering properly so I had to feed in a touch of opposite rudder during AutoTune…

I had the same problem with my 9xr pro, even replaced one of the gimbals and the problem kinda got away. But I had to calibrate it almost every time as it was drifting even with the new gimbal, not so much as the first one but the problem is still there.
The final resolution was to buy a Taranis X9D plus and all my problems went away.

So i have tried the 9xr pro on another copter getting the same yawing result, so it is definitly the TX sending wrong/noisy signals to the model.
I will recive a X9D in the next days and will switch to this tx instead of the 9xpro and will post results after some tests, but i think this thread can be closed/marked as solved.

thanks for your hints