DJI Drone Help Forum

DJI Spark Fly away (lost)


Hi All,

Yesterday I lost complete control of my DJI Spark leading to the loss of the aircraft. It was at a height of approx 30m when it started flying off in a straight direction away into thick forest, completely unresponsive to the controller. I was unable to find the aircraft as there was no GPS signal recorded. I do have Care refresh for this but understand you need to recover the aircraft however this was impossible to find given the terrain involved and no GPS location recorded. I would really appreciate any help someone can give me on this.

The flight log can be found here:

Please let me know if you need any further information


There may be some GPS info in the .DAT recorded on the tablet. Here is some info about how to retrieve that .DAT


The tablet .DAT file name will have “FLY162.” in it.


Thank for the reply Bud, yep there were 6 .DAT files that I have zipped and put into a file here:

…just not sure how to read these



I’m still looking at this but I think your Spark may be here
153.284285 -28.2115213
or here
153.2843906 -28.2114191

It would seem that the Spark remained on and was connected intermittently to the RC. That’s why there were 5 .DATs recorded on the tablet. The Spark had just enough GPS sats to obtain a solution, but not a very good one. Looks like you’re in for some serious hiking.




Data stream says it took off with ZERO satellites and likey just drifted away in the wind . . it only ever saw 5-6 satellites a min or so later . . so those locations are a start but can’t be very accurate. Did it initiate it’s own takeoff un-commanded or did you not realize it was in ATTI sometimes referred to as “Drift” mode when you gave it the ok to takeoff. Pretty much needle in a haystack in those woods . . good Luck . . hope you find it.


The Spark was connected intermittently to the RC which caused 5 separate, sequential .DAT log files to be produced. The 2nd .DAT shows the motors stopping. Presumably, the location didn’t change after that.

At that time the gps count was 6 sats (2 GPS sats and 4 GLNAS sats). The dilution of precision numbers weren’t awful.

In the 4th .DAT there was a brief period of 3 sats and the GPS module produced a location very close to that of the 2nd .DAT.

Since these two solutions are very close they a probably very close to the actual location.


The Spark wasn’t in ATTI mode until 43 secs after launch. It was in GPS_ATTI mode which is somewhat of a misnomer. What this means is that the Spark is trying to navigate. But, GPS wan’t being used. Instead the Vision system was being used for the navigation.


So what was actually happening between 0 and 60 sec? . . . looks like there is a reasonable chance to locate if there was location agreement from two sources.


I didn’t look much at the first 60 secs. The two .DAT log file that I mentioned were after the Spark had landed but there was intermittent connection with the RC.


Wow thanks a lot for all the help guys, really appreciate the details. Yea I did realise it was in ATTi mode and probably wasnt the best idea given the area I was flying it in. So is there good indication of remote loss while in there air? I thought I had a compass error flash up on screen at one stage but the logs don’t seem to suggest this? So the primary and secondary co-ordinates would be the best to go on? …haha even so the mission to get this would be insane, it’s a 2 hour drive away and then an 18km hike to get in and out, not to mention I’d have to use map and compass on these co-ordinates to find (without sufficient GPS signal) :S


But, it wasn’t in ATTI mode as I mentioned above. You did, however, launch with the GPS level at 0 (seen on the RC display) . Here is an example of the GPS level at 4.GPS

There were no indications of a compass error.


OK, then, I thought it was all about finding the Spark.

I think probably what happened was that the Spark was using the Vision system to navigate and couldn’t do very well looking down on the tree tops.

At time 67 secs it looks like the Spark ran into something. I’m guessing it got tangled up in a tree. Here is the acclerometer and gyro data. The encounter lasted about 7 secs.

The motors were commanded off at time 70 secs because the Spark detected that it had rolled over. This can be seen in the eventLog stream
64.162 : 6984 [L-FLYMODE]CTRL rc none mission!
64.562 : 7004 [L-FLYMODE][Ctrl<1>] REQ_RC_NORMAL ATTI_HOLD ctrl_gps_atti
66.282 : 7090 [L-FLYMODE][Ctrl<1>] REQ_RC_NORMAL ATTI ctrl_atti
67.702 : 7161 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
67.722 : 7162 [L-RC]craft ctrl failed!!!
67.782 : 7165 [L-FDI][CTRL]: fault off, tilt_ctrl_fail


Thanks Bud, so would there be a chance it’s stuck in a tree then too? I was also wondering about the Max Height reached message when I’m very sure it was no where near 140m up.



Looks to me like it’s stuck at 8 meters higher than the launch point.

I don’t know what the Max Height Reached messages are about.


Hmm there was elevated terrain in that direction I guess so it still might be on ground. Would the Maximum Flight Altitude Reached error (which restricted me going higher) be due to the drone using the Vision Position System?


Unsurprisingly DJI have said this is not covered under the Care package until I find it, so looks like I’m going for a wee hike. Sorry to ask for further help but can you let me know what the GPS co-ordinates for the remote (Pilot/Tablet) from the picture in your post from 23/09 11:30pm.



The tablet/pilot location comes from the .txt log file. In Google Earth image is -28.212246,153.28567

At the very end of the .txt there is another tablet/pilot location which is

Apparently, the poor GPS reception affected the tablet’s GPS receiver as well.



Looking forward to hearing if the ground search was successful or not.