OK, thanks all. I've updated the links again with the 2 records that were near the time he gave me, hopefully one of these will be the right flight. If you look at the flight above, the flight in question basically went from "H" to "A" where it went crazy then basically flew north and crashed into a tree.
Thanks again!
FLY013 was the one. It was pretty much as you described, but with a couple of significant differences. The first graph shows the cause of the problem, I think - a very large compass error around 80 seconds as illustrated by the sudden divergence of yaw vs. magyaw. Both compasses reported bad data at 85 seconds.
85.492 : 395039785 : 4132 [L-FDI][FDI][COMPASS(0)] is outlier(fault)
85.492 : 395039950 : 4132 [L-FDI][FDI][COMPASS(1)] is outlier(fault)
87.074 : 402161265 : 4211 [L-FMU/LED]action changed. compass out of range
0)
88.475 : 408463710 : 4281 [L-FDI]NS(0) FUSION(0): fault on , magn_heading_err_large
The GPS lock was fine and interestingly, it reports not flipping into ATTI mode (until after impact). Also, it appears to have continued to hover for another 20 seconds or so befoer starting to move north and east (which looks like ATTI drift even though it was still reporting P-GPS), causing the collision with the tree at around 113 seconds. That loss of horizontal control was reported at 106 seconds followed by a bunch of other problems in quick succession:
106.774 : 490811221 : 5196 [L-FDI][CTRL]: fault on , horiz_ctrl_fail
107.574 : 494411776 : 5236 [L-FDI]NS(0) FUSION(0): fault off, magn_heading_err_large
109.930 : 505011057 : 5354 [L-FDI]ns req:fdi,0to0,reason:fusion.gps_yaw_err,result:fail for same index
111.074 : 510162261 : 5411 [L-FMU/LED]action changed. imu error:ns_abnormal(3)
111.134 : 510429569 : 5414 [L-FLYMODE][Ctrl<1>] REQ_RC_NORMAL ATTI ctrl_atti
111.154 : 510520807 : 5415 [L-FDI][CTRL]: fault off, horiz_ctrl_fail
113.534 : 521232526 : 5534 [L-FDI][CTRL]: fault on , impact_in_air
113.552 : 521312818 : 5535 [L-RC]craft ctrl failed!!!
113.588 : 521474004 : 5536 [L-FDI]NS(0) GYRO(0): fault on , stuck
But there is nothing to indicate that control was actually lost. The second graph shows stick inputs and aircraft response. Apart from the rudder input at 80 seconds that triggered the compass error, there were no elevator or aileron stick inputs after that to attempt to control it. There was a brief throttle down at 87 seconds that did produce a slight altitude drop, and a rudder input at 99 seconds that the aircraft responded to appropriately.
In summary - that looks like a significant sensor failure, and I'd hope that DJI would replace it.
