Can anybody help analysing the cause of this crash?

The copter started a test mission without any troubles and after reaching a waypoint, he started to yaw constantly but continued to follow the mission path.

After a couple of seconds the pilot switched out of AUTO mode into PosHold and then down to STAB, but the copter kept on spinning. The resulting oscillation of trying to get the copter stable and dealing with the ongoing spin resulted in a crash.

When I looked at the log-files, I can see the actual yaw following the desired yaw nicely. Until this very certain waypoint where the desired yaw goes up to 360, stays there for 3 seconds and then drops to 0 and stays there - all still during AUTO mode. When the desired yaw goes up to 360, the copter started its constant yaw and did not stop unitil crashed.

We are aware of the z-vibrations being not super nice, but I am pretty sure that this is not the cause of this behaviour. The copter holds it's altitude pretty nicely.

Frame type is a X8.

Any ides or suggestions?

Any help is greatly appreciated!

Regards

Marc

Views: 3877

Attachments:

Reply to This

Replies to This Discussion

Cala, thanks for tuning in!

I did not fly nor build the copter and the pilot couldn't remember exaclty when he switched back to manual mode or how long he tried to countersteer against the yaw.

Can you tell me more about the Clips as it seems that I am not up to date with the latest logging/analysis possibilities?

The guys managed to reduce the z-vibes to a fifth of its original magnitude, so this shouldnt be a problem any more.

Check, in case, if you don't have something bad fixed on your frame; can you share some photos?

Here you have info about clips http://copter.ardupilot.com/wiki/common-measuring-vibration/

Dear Community, hello Cala,

as i am the user of the system i could take a few photos.

Please find them attached to this post.

Greetings

Alex

Marc:

I have a large Y6 and had a similar crash from uncontrollable yaw that started during a mission that started well.  After reviewing my log files I could not find a good reason as all data showed the controller attempting to stop the yaw.  In stabilize I could not effect the yaw.  In the end I concluded that some recent work to the motor mounts resulted in NOT doing the final tighting of one motor mount thru bolts, I expect the entire motor mount, with top and bottom motors and props, rotated on the arm during the flight and caused the uncontrolled yaw.  Its been months in a rebuild, all my fault.  Now I check the motor mounts every time I fly.

Is there any chance the mounting blocks got loose somehow?

At work so can't look at the logs, but looking at the pics Cala posted, I had something similar (which I still haven't resolved). As Cala said, the vibes are suspicious. Also try looking at the EKF mag innovations. There's a howto on the wiki on how to analyze EKF log data. http://dev.ardupilot.com/wiki/extended-kalman-filter/#interpreting_...

@Harry,

loose mounting block can be easily detected in pre-flight diagnostics

if you FFT Acc outputs amplitude and frequency don't match amplitude and frequency of airframe resonant vibrations.

Darius, checking the motor blocks is now a regular routine.  A difficult and expensive lesson I will never forget.

How do you compare the resonant frequency of the frame to the Acc outputs in pre-flight diagnostics?

In my case the motor blocks rotated after 3 minutes into the flight.  Pre-flight diagnostics showed everything was good to go.  Old fashion grab and twist is the only pre-flight test I have come up with. 

Darius is talking about a hypothetical software solution to look at accelerometer data to try to ferret out loose mountings.

I prefer a physical inspection, as you did Harry.  You're saying the mounting blocks became loose during flight?  Are you using any kind of threadlocker on the screws?

Typically, every nut, bolt, and thread have some type of nail glue or threadlocker.

This one time I had disassembled my motor mounts, re-assembled, and perhaps tighten all but one set of blocks?  It's a question, as the damage was substantial.  Booms and four motors where broken or missing.  I was very lucky to recover and use all the expensive parts.  Its only my guess that this caused my crash.  Upon impact, all three battery connectors where pulled out, so no log file was recorded, however the telemetry log indicated the flight controller was taking the correct action to control the yaw, but had no effect.  That leads me to conclude this was a physical problem.  It makes me sick to think I was so stupid to allow this to happen.  Now I grab and twist every motor mount before every flight.

I have had similar result after motor mount twisted in flight. Don't forget the whole arm can twist too.  I've found single tube arms to be such a pita, I won't use them anymore. Without a locking hole/groove at both ends you can't be confident not to get twist but putting those in carbon significantly weakens the arm.

Darius, You couldn't find your ass with  your own two hands and a flashlight.  I've asked you to profied your credentials but you have never responded with them.  I don't pretend to be perfect but you do yet you won't provide any of your abilities.   I therefore must assume you are a troll and only exist to be a thorn in the side of those who seek the truth.  

Why Darius hasn't been banned from this forum is beyond me.  I'm as much for Freedom as the next guy.  I'd die for the right of anyone to speak.  However, Darius hasn't added anything worthwhile to any discussion.  That he still has an account is a miracle that only God himself can explain!

Hi Marc,

This is definitely a mechanical issue of some sort. As has been stated already, the copter is doing everything it can to fight the yaw but is unable to do so. This is why it doesn't matter what mode it is in. The controller has been deliberately setup to sacrifice yaw in order to maintain roll and pitch in these situations. This lets you land safely and correct the problem.

I can't point to which arm is twisted because you were in a dynamic flight state at the time. However, a twisted arm is the only thing I can think of that would cause this issue. A single lost motor should be clear in the logs and should not result in lost yaw authority (as you demonstrated).

I have been fooled by this myself once as it doesn't take much of a twist to overcome the limited yaw control of quads.

Reply to Discussion

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service