Hexacopter crash at beginning of roll-autotune after secessfull yaw-autotune

I have a tarot 690 hexacopter with multistar 3508-640 brushless motors, 30A Afro ESCs with blHeli 14.4 on 4S 8000 mAh 30C and a 3DR Pixhawk with copter firmware 3.3.2

I was flying the hexacopter for some time with a GoPro to take some areal shots.
Then I switched to a larger samsung camera and a bigger gimbal. So I had to redo the autotune process.

With the GoPro the ESCs had the simonK firmware installed. But I had to switch one of the ESCs and didn’t know the firmware version and settings of the other ESCs. And because I had great results with BlHeli on my mini quad I installed BlHeli on all six ESCs with the same settings.

Today I went out flying and enabled autotune (yaw-axis) which worked perfectly. I also had time to fly a bit around after it because the battery was at about 50%.
After that I drove home, charged the battery, plugged in the Pixhawk to the computer and changed the autotune axis from yaw to roll.
Drove back to my flying spot, started and switched to autotune (in attitude mode like it should be) but nothing happended. So I landed, disarmed, armed and started again. Now autotune started on the roll-axis.
But after the first or second move the copter tilted to the left (roll) until it hit the groud. Then it jumped back in the air a few cm but leveled as it should be. In that moment I switched to stabilized mode and cut the throttle.

Looked if it was damaged (just a few fixable bends at the gimbal) and tried to start again. So I started in stabilized mode an wanted to test roll by putting in some left and right roll input and the copter immediately tilted to the left side and hit the ground again.

So I went home again and had a look at the log files (attached). The “auto analysis” of the first flight was fine (yaw-autotune), the auto analysis of the second flight (roll-autotune and crash) got me an IMU mismatch warning and the last flight (test after crash) analysis got me no warning despite showing the same crash behaviour as the second flight.

Is it normal that some signals (Gyro z for example) have small “gaps” where no data is available? Is that because of the logging or could that be a problem?

I hope someone can help me find the problem.

Edit: I can’t upload the files so I uploaded them to dropbox:

Hiya, looks like you’ve got a fault with your number 2 motor or ESC (middle left if you’re using hex X layout, middle back if you’re using hex + layout - I’m guessing hex X as it looks like it’s roll that triggers the issue). It could either be a dodgy motor or connection, or most likely it’s an ESC desync. You’re not using blheli damping light mode are you?

Yes, I’m using BlHeli with damped light on.
Is the damped light mode more likely to get sync problems with motors?

Was thinking about goingt back to SimonK.

But I don’t know which version of SimonK was installed before, because I don’t know how to read the firmware version of SimonK installed on a the ESC.

Yes absolutely, it seems to trigger desyncs in lower kv motors. I have 3508-580kv motors which flew dozens of flights perfectly, the first flight with damped light and it desynced and nosedived.
Blheli is better than simonk with lower kv motors as the throttle response is smoother - simonk was designed for racing so the throttle is much sharper, as I understand it.
I’m absolutely no expert but to me the logs show a clear desync or else a mechanical problem. Do check your number two motor and ESC as that was the cause both times.

Turn off damped light and try again. You can try different demag and timings as well but I never found they made much difference with me.

Turned damped light off and tried it again. And again it crashed in stabalized mode.

BUT I tried again and could complete the auto tune process without any problem. So there is no syncing issue anymore!

After auto tune was complete it had tuned down the PIDs drastically and it was nearly unflyable. Roll was all over the place and had completely lost its “locked in” feeling. But I wasn’t able to get it to crash again. “Tied” it till the battery was empty.

The flying weight is absolutely maxed out with the Samsung camera and I have a theory why it crashed despite the syncing issue has been solved.
On a hexacopter there are TWO motors in the front and in the back to manage the pitch-axis but only ONE motor on either side to control the roll-axis. So I think that with my fast roll inputs in stabilized mode it hadn’t enough power to stabilize roll and crashed.
When I activated autotune (hovering wasn’t an issue) it tuned down the PIDs so much that it got totally soft/wobbely/spngy but was able to hold it in the air without falling from the sky. Because with the low PIDs it wasn’t possible to accelerate the hexacopter so fast on the roll axis that it wasn’t able to stop the movement.

Until now I flew with 12x4.5 propelllers and now I switched to 13x4. Hopefully that is enought to get enough power to fly it like it should. The motors will have no problem with it as they came down completely coll with the 12x4.5 and others fly those motors with 14" propellers.

I’ll report back when I tried it.