Plane crashed vertically at top speed during a mission in auto mode

Hello All,
I will appreciate your help with this issue, as I have no idea what could be the reason for the crash. The plane WAS a flying wing NINOX from Bormatec running arduplane 3.7. It was flying perfeclty the day before, flying many times the same planned mission in auto mode, with only one minor problem: it reduced too much throttle and lost too much altitude at the start of the landing, but it recovery and landed as planned. So I made just one change: TRIM_TRHOTTLE from 45% (default) to 75%, as it was clear that it need a lot of throttle to level flying.

With this single change, I tried today the same AUTO mission but it behaved badly, too much speed (20 m/s instead of the configured 12 m/s) and missed most of the waypoints (I got frequently the message “resseting to the previous waypoint”).
You can see this behaviour in the attached telemetry and dataflash logs at 18…22%. The plane crashed gently, but I notice in the telemetry log that the the target altitude was not always met with large errors (+10 meters) sometimes, and, previous to this gently crash, altitude reading was negative (-8 m, which is not possible).

So, I decided to change the TRIM_THROTTLE parameter back to 50%, and flew it again with the same mission (starts at 98.3% in the telemetry log). Once It got the target takeoff altitude of 10 meters, instead of flying to the WP 2 ahead, turned back to me and went down almost vertically at top speed and crashed violently.

I have no idea what could be the reason. Any help will be very much appreciated because, sadly, I am losing my faith in arduplane :frowning: I really hope is was my fault.

Crash is at 98.3 % of telemetry log and at the end of the dataflash log.
Thank you in advance!

ninox 2017-04-14 12-05-27 estrellado.tlog (1.8 MB)
Download link for dataflash log:

Hello to all,
I am still struggling to find the cause of this crash, and I have found the following in the dataflash logs that could give a clue. I will appreciate your help with this.
The problem was that right after the take off , with the WP2 ahead, the plane turned back to me and crash almost vertically at top speed.
If I plot the “mechanical failure” params (intended and real roll & pitch):

It seems that the desired and real roll started to diverge around line 428k, before crash at line 430k. If the sensors/accelerometers parameters are plotted, It seems that around line 428k there is a “glitch” in all three axes:

The thing is that the horn that connect the servo to the aileron of the right wing was broken after the crash, but I supposed that was due to the accident. However, It could got broken in flight and caused the loss of control, signaled by the glitch in the accelerometer signal. Could it be possible?
Thanks in advance for your help,

I’m no log expert but looking at Roll/des.Roll v RCout1 and 2 it does look like a mechanical failure as you have surmised.

The des.Roll and Roll diverge and RCout maxes out in an attempt to correct.

My guess is, yes, you lost a linkage.

Not sure about why the turn around though.

Maybe someone with more insight than me can help.

Thank you very much for looking into it!
I didn’t think of looking the RC outputs, if the autopilot is trying to correct the attitude, the theory of a broken linkage gains more points.
the plane is a flying wing with no rudder, so the lost of one aileron could also explain the change of trajectory, I think.
Thankyou for your help,