Crash after switching to loiter mode. Help analyze

Hi.
Could somebody help me analyze my plane log. This was the third time I flew this plane and the first two flight where mostly trimming and finding a good CG, but I succesfully made one circle in loiter mode, and some 50 meters in fbwa mode. Today after I switched to Loiter the plane started to turn right, added throttle and never stopped turning right, so when it was about 90 degrees I switched to manual but couldn’t save it and crashed nose down into the ground.
I’m not an expert it logs. All I could find is maybe in ATT.Roll and ATT.DesRoll. In the graph looks like the DesRoll stopped at 50, but Roll went to 150. Odd.

Aruplane 3.9.3 on pixhawk.
Upload to forum didn’t work, link to the log:
https://drive.google.com/file/d/1YvL3MZd9weTwbdsk8RlFjMiwsdbH19zL/view?usp=sharing

Thanks!

Forgot to tell it’s a V-tail plane, mini talon.

I guess that except in manual mode the ailerons are moving into wrong direction. See http://ardupilot.org/plane/docs/starting-up-and-calibrating-arduplane.html#check-servo-movements about #1 cause of crashes .

Rolf

Really? But I made a flight before that and it worked.
I didn’t check pre-flight in fbwa mode for this flight. I still don’t understand why it would change between flights. Actually the only change I made was to loiter radius from 60 to 40 meters.

I now started thinking that it’s possible that I accidentally hit the reverse checkbox in Mission planner because I was thinking on trimming the elevator from there but then decided to not to. It’s possible that I didn’t see that I clicked on it because of the sun shining on the laptop.

I can’t extract params from the previous tlog flight and I don’t remeber if it had to be reversed or not. Is there any way to extract params from the dataflash bin file?

Oh, Rolf forgot to thank you. I will check when I hotglue the plane together. Thanks!

On a side note. Mission planner should have a Save button on the servo page, that way it wouldn’t be possible to accidentally reverse servos just with one click.

I managed to extract the good flight params by updating mission planner. Everything was ok, the aileron was already reversed in the first flight when everything worked.

I made a diff of all the params that changed between a good flight and bad flight:

good flight bad flight
COMPASS_DEC 0.06090735 COMPASS_DEC 0
GND_ABS_PRESS 51015.58 GND_ABS_PRESS 50394.17
INS_GYR2OFFS_X 0.01245254 INS_GYR2OFFS_X 0.009244195
INS_GYR2OFFS_Y -0.05279394 INS_GYR2OFFS_Y -0.03473469
INS_GYR2OFFS_Z 0.003622743 INS_GYR2OFFS_Z 0.001236723
INS_GYROFFS_X 0.005003656 INS_GYROFFS_X 0.005379174
INS_GYROFFS_Y 0.0002330209 INS_GYROFFS_Y -0.001001616
INS_GYROFFS_Z 0.0008673746 INS_GYROFFS_Z -0.003074314
KFF_RDDRMIX 0.5 KFF_RDDRMIX 0.6
SR1_EXTRA1 4 SR1_EXTRA1 10
SR1_EXTRA2 4 SR1_EXTRA2 10
SR1_EXTRA3 2 SR1_EXTRA3 3
SR1_PARAMS 10 SR1_PARAMS 0
SR1_POSITION 2 SR1_POSITION 3
STAT_BOOTCNT 18 STAT_BOOTCNT 25
STAT_FLTTIME 168 STAT_FLTTIME 774
STAT_RUNTIME 16897 STAT_RUNTIME 28482
SYS_NUM_RESETS 27 SYS_NUM_RESETS 34
WP_LOITER_RAD 60 WP_LOITER_RAD 40

I can’t understand why this happened. Of course it was my fault with the preflight check.

Take a look at X-axis qyro and aileron servo output:

The gyro reacts to aileron movement in manual mode and loiter mode in opposite directions.

Thanks. But from what I see the gyro didn’t even react, it went more or less a straight line. Could be the gyro at fault? The previous flight went ok like I said.

Just wanted to update that I probably found out the cause of the crash. It was probably a combination of the plane pitching down a little when giving full throttle and not having set enough climb rate in auto modes. When I set loiter to only 40m the plane would go full throttle and full climb, but that wasn’t enought to keep the height. The full throttle picthing down mad things worse, probably because of a bad motor mount angle.