Spark initiated a forced landing - analysis assistance required

[Yanickma]

Hello all,

Today I freaked out a bit because I almost had an accident with the little Sparky. I d like to request some assistance and feedback from you to make sure I understand what happened and to avoid this to happen again.

Still new with the Spark, only one month or so of flying it. I was on the edge of a cliff, took off from palm and pushed the drone up and backward (like a dronie shot but done manually with the RC).

The cliff was high (couple of hundred meters deep below filled with tropical forest trees. when i finished the shot i was hovering there when suddenly the RC started to emit regular beep beep and the drone started to descend. I didnt panic but started to freak out a bit since it was not responding to go up. I still had control to move one side to another or forward backward but no control on the vertical axe. Luckily enough i was not too far from the edge of the cliff so I was able to take it back just on time to land beside me. If not, it would have slowly gone down the precipice and crash into the trees several hundred of meters below.

Important fact, once i flew up and finished my shot, a sudden fog went down and reached the drone in a matter of seconds ( i m in Costa Rica and rain forest weather can change in no time, it went from clear to foggy in less than 30 sec)

Reading around I m 90 % sure that this sudden fog triggered the force landing. But I would still like to have your opinions.

I just want to make sure I fully understand what happened and all the variables involved.

Mostly, I d like to confirm, it is OK to take off a Spark on the edge of a cliff and fly it over a canyon, I mean , that would be kind of lame if the drone could not handle this sudden change of ground reference (ground level at the edge of the cliff vs ground level at the bottom of the canyon). I dont think it would go in force landing just because of that right?

Also once it landed and I lowered my eyes back at the app (during the emergency I locked my eyes on the drone only) there was a red compass error message. Not sure if it s a consequence of the issue or a cause.

I was 90 km away from home, drone did not ask for compass calibration and I flew 7 other times this day without issue.

My thoughts are the issue was primarily caused by the sudden fog, and maybe I did wrong by not calibrating the compass even if the drone was not asking it, but I ll appreciate your input.

I added the flight log:

DJI Flight Log Viewer - PhantomHelp.com

I also downloaded the black box and flight controller DAT file, if needed you can let me know.

I appreciate all your good help!

P.D:

This is a snapshot of the video around when the problem started. Please dont bash me too hard, as mentioned, here the fog can be tricky and start in a matter of seconds

 

[r-ap]

Hi Yanickma,

in my opinion, forced landing and compass error are two separate issues.
A few weeks ago, I‘ve tested sparks behavior in light fog. I started up to 10m (all in open area) an tried to circle around a flagpole. Suddenly RC began beeping and Spark started forced landing. No possibility to cut the descent, must catch Spark by hand.
Probably the downward sensors of Spark were disturbed by the fog (no compass error for me)


r-ap

 

[Yanickma]

Hello r-ap

Yes I think the fog is a more possible cause.

Adding the DAT file run thru DatCon

Flight was around 5:07 pm local time on 10/21

 

[Yanickma]

Something else I also see in the log, starting at 35 sec, the VPS alt goes from 19 ft to 0 ft, and it will mainly remain between 0 and 2 ft for all the rest of the flight while the IMU alt is much higher.

Not still sure to understand the difference between both. Is it that IMU is alt from barometer and VPS is alt from some kind of visual?

If so that would reenforce the fog confusion theory right?

Thanks!

 

[Kilrah]

VPS is the combination of the bottom camera and infrared ground height sensor. There is already a big thread on here warning that fog being reflective makes the IR sensor read 0/low value for ground height which means as soon as you try to descend the aircraft will think you want to land it and cause what you experienced.

 

[d0ppler]

I've descended through clouds/fog with my Spark. I had full control over the aircraft, but my phone was beeping all the time telling me altimeter is 0.3m, even though it was 50m up in the air.
You can see the footage here:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

 

[Yanickma]

I've descended through clouds/fog with my Spark. I had full control over the aircraft, but my phone was beeping all the time telling me altimeter is 0.3m, even though it was 50m up in the air.
You can see the footage here:

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

Amazing vid!! But the spark never wanted to start a forced landing because of that fog?? Mine was much less foggy and it complained right away and started to go down with no return. Anyway for me never ever flying in fog i learned my lesson!!

 

[Kilrah]

I've descended through clouds/fog with my Spark.

You basically got lucky, there have been several reports of people not having such a nice outcome. Just keep it in mind...

 

[d0ppler]

Yeah, I will most likely never do it again, since my spark was soaked in water after the flight.

 

[BudWalker]

Hello r-ap

Yes I think the fog is a more possible cause.

Adding the DAT file run thru DatCon

Flight was around 5:07 pm local time on 10/21

I looked at the .csv generated from the .DAT by DatCon. The FORCE_LANDING was caused by vpsHeight < 0.5 AND Throttle == -10000. It thought that it was within 0.5 meters of the ground and then you gave it full back throttle.



When looking at the .txt when vpsHeight == 0.0 that means that vpsHeight doesn't have a value. CsvView/DatCon presents this better as can be seen in the above. When vpsHeight doesn't have a value it's not shown.

 

[Yanickma]

I looked at the .csv generated from the .DAT by DatCon. The FORCE_LANDING was caused by vpsHeight < 0.5 AND Throttle == -10000. It thought that it was within 0.5 meters of the ground and then you gave it full back throttle.
View attachment 1727
View attachment 1728

When looking at the .txt when vpsHeight == 0.0 that means that vpsHeight doesn't have a value. CsvView/DatCon presents this better as can be seen in the above. When vpsHeight doesn't have a value it's not shown.

Thanks a lot for the indepth analysis.

According to this it still makes sense the fog caused that messed up vps value?

It s true i was giving full back throttle since i was recording a dronie like motion backward.

 

[BudWalker]

Thanks a lot for the indepth analysis.

According to this it still makes sense the fog caused that messed up vps value?

It s true i was giving full back throttle since i was recording a dronie like motion backward.

Yes, my guess is that it was the fog that caused an incorrect vpsHeight value. But, it's the combination of vpsHeight < 0.5 AND full negative throttle which causes the FORCED_LANDING. So, if you had avoided full negative throttle then the incident would not have happened. You can also avoid this situation by setting that has something to do with Landing. I can't ever remember what exactly this is so I'm going to dial 1-800 @msinger who will know.

You referred to back throttle as causing motion backward. This isn't correct. The throttle causes altitude change.

 

[Kilrah]

Unfortunately the "landing protection" cannot be disabled on the Spark.

 

[Yanickma]

Yes, my guess is that it was the fog that caused an incorrect vpsHeight value. But, it's the combination of vpsHeight < 0.5 AND full negative throttle which causes the FORCED_LANDING. So, if you had avoided full negative throttle then the incident would not have happened. You can also avoid this situation by setting that has something to do with Landing. I can't ever remember what exactly this is so I'm going to dial 1-800 @msinger who will know.

You referred to back throttle as causing motion backward. This isn't correct. The throttle causes altitude change.

Thank you very much....I thought by back throttle you meant going backward....so back throttle means going up? Or going down? The "back" part is what confuses me. Now you said negative throttle so i would assume it s pushing down but if you confirm i would appreciate.

 

[BudWalker]

Thank you very much....I thought by back throttle you meant going backward....so back throttle means going up? Or going down? The "back" part is what confuses me. Now you said negative throttle so i would assume it s pushing down but if you confirm i would appreciate.

Throttle (left stick) controls altitude. Pull it back (towards you) is negative throttle and will cause the AC to descend. Push it forward (away from you) is positive throttle and will cause the AC to ascend. Your incident happened when vpsHeight < 0.5 AND then you pulled the left stick all the away back which was full negative throttle.

Elevator (right stick) controls speed, both forward and backward.

 

[Yanickma]

Throttle (left stick) controls altitude. Pull it back (towards you) is negative throttle and will cause the AC to descend. Push it forward (away from you) is positive throttle and will cause the AC to ascend. Your incident happened when vpsHeight < 0.5 AND then you pulled the left stick all the away back which was full negative throttle.

Elevator (right stick) controls speed, both forward and backward.

Good! Thank you for the explanation sorry for missing some part of the jargon , that s part of thr learning. Yes, i dont remmber if i tried to lower it down just before the rc beeping started or just after it started, but yes my first instinct was to start a descent when the problem initiated.

I ve learned a lot with this incident and all the good analysis everyone provided