Crash on Autotune (version 3.1.2)

Hello folks,

today I experienced a crash on Autotune when I was trying different props (changed from 13x4 to 12x4.5). The hex did fly not so bad and I engaged autotune. The pitch axis was very quickly tuned and when it comes to the roll axis id did 1 or 2 twitches and then the hexacopter rolled over the roll-axis (sideways) and fell on the ground (from about 4m of height).

Damages so far: All props broken, frame cover broken, 1 motor support broken as well as the motor. :frowning:

I tried to analyze the problem when reading out the log (Standard+IMU) and I cannot find the cause for the crash. Vibration levels are good, compass-mot was done, accelerometers and compass correctly aligned as well as RC calibrated and ESC calibrated. There was also no power issue, I flew with 2 batteries each 5000 mAh and they had both perfectly 12.3 volts after the crash. Vcc was stable slightly above 5 volts since I am using an external UBEC on the input side of the APM.

It sees to me that the APM suddenly raised throttle to 100% and then corrected it by turning throttle down to THR_MIN. At this point the hex rolled over and I had no chance to react by switching back to Stabilize.

If anyone could help me with the log?

Kind regards!

Sven

Additional information:
The hexa’s weight is around 2.6 kg, motor to motor the frame is 800 mm in diameter. Motors are T-Motor MT2216-12 (800kv) with 12x4.5 glass-fiber reinforced nylon props. ESC are RC-Timer 30A P/N-FET with custom SimonK firmware.

The thrust to weight ratio is around 2.2, so not over and not underpowered. The hexa flew very well the last 10 flights but I wanted to go back to the other props because the 13x4 did not gave enough thrust to compensate altitude loss when doing fast flights in althold mode.

GPS: 3DR LEA-6H
Radio: 3DR telemetry 433 MHz

All ESC and motor connections were soldered. I’ve checked all connections and I can exclude a connection failure. The defective motor died at the collision (bent axis) and showed no play in the bearings, no smoke not heat or whatever else. I usually do an extensive pre-flight check on this as well as the props and the cables and all was perfect before flight.

Sven

None has an idea?

Greetings!

Sven

Yes, plenty people here have an idea, but gets tired of answering same question :slight_smile:
If you searched for “crash autotune” , you’d find plenty of answers - telling about motor sync problems.
it’s about your motors vs ESC vs firmware and timing setting of ESC.

The motors do not have any sync problems. I compiled the SimonK firmware by myself and did days of extensive testing to ensure the motors do not have any sync problems.

The motors do not have a high pole count, they have standard 14 poles and have an apopriate timing advance set. I have tested the motors under load with a programmed switch on my remote which rises throttle from 20 percent to 100 percent by flicking that switch and there were no signs of sync losses when quickly switching between these two and (and other combinations) of throttle.

If you would have taken a look into my attached log file you would have seen that autotune raised the throttle to 100% for some reason and then reduced it to THR_MIN and let it there. After this the hex started to roll over, so the crash is directly caused by the throttle output of the APM. Besides this
I have done autotune with these engines around 20 times before without any problems.

I do not blame the APM developers for any errors that could be in the code and leading directly to a crash although I have lost several hundreds of dollars in crashes that were not all my fault alone but I expect a help here in the support forums when I do report an issue. Instead of that I get arrogant answers that my motors have sync problems


Andre, the problem was and is that the APM reduced throttle to minimum for some reason and did not raise the throttle up again. The crash is the direct consequence of this.

Sven

If we can agree that Rolls falls a lot, without RollIn causing it:
Please zoom in on the graph, already at line 10400 you have an unusual pitch of -5deg , and ThrOut is not unusually high.
Then Throut climbs rapidly (remeber this is the total throut, you did not log motors) - and this is not uncommon as it commands more throttle (most likely very much to the falling motor) .
Then: at line 10428 , it “cuts throttle” - (actually, it reduces it rapidly) - why ? - because at that point - as logged at 10430, your pitch is -45.67 degree. It would not make any sense applying more trust, if it worked, that would make it reach extreme speed, so APM is reducing the thrust - it actually cuts the thrust when pitch is -140 degree 
 if it did not, it would fly hard into ground.

So OK, I cannot be sure about the “sync” - (yes, it’s it was a hasty suggestion), but I still see a hardware failure somewhere between PWM out and propeller.

Thanks for your reply Andre and your time to have a look into the log!

Ok, I will check again all connections and the ESCs of the side the hex rolled over to. When I get all parts to repair the hex I will do extensive tests on this.

Please wait for my reply.

Regards!

Sven

Hello André,

I am coming back to this topic: Could it be that my seen behaviour is caused by the stability patch bug in AC-3.1.2 that caused the motors to shut down when high throttle and high roll is applied simultaneously?

I think Autotune could trigger that condition


I have tested all ESCs and motors and can’t find any bad solder joints (I know how to solder, I am a ham radio operator) or defective ESC. I even cooled down / heated up every single ESC while running on my test bench to exclude any bad solder joints on the ESCs.

I really cannot find the cause for the last crash and this makes me nervous


Regards!

Sven

were all engines and propellers still attached after the crash ? (see ATT roll)
It could be also a propeller screw be dropped out or something


Hello Pomaroli,

Many thanks for your answer! All props
well the rest of them
were still firmly attached to the motors. I am using CW and CCW threaded prop adapters so that the nuts cannot become loose.

In the meantime I can exclude 100% any mechanical failure and almost 100% an electrical failure. The only thing what is left is that stabilize patch bug


Kind regards!

Sven

Hi Sven,
Did you find the reason? I face same issue when auto tune. When auto tune enter step two, ThrOut became 100% suddenly, the mulit-copter yaw and crash.
I fixed the leg then tried to flight it again, no any problem found. Don’t know should I try auto tune again.
( Flight log attached )

Best regards,
Bellergy