Spark Ground Addiction

[Adrian]

My Spark has a ground addiction it keep hurlying to the ground whenever it feels like in sports mode. I posted this before but I didn't have the logs, now I was able to get the logs and I was hoping someone could help me find out why this keeps on happening.

Dropbox - DJI_ASSISTANT_EXPORT_FILE_2017-08-19_11-27-32.DAT

thanks,
-Adrian

 

[Dronason]

At first, the compass seems to work nice when in flight at some altitude.
for flight at some altitude

When you were in Sport mode (before the crash), you first were full elevator forward, speed increased up to 9 m/s. You started to release the elevator and at same time put rudder to maximum.
So at same time it was braking it had to rotate it reached a pitch angle of 65° and roll angle of 73°.


The Right Back motor get's to its maximum and was not able to give more.


So I would say that IMO that you discover where are the flight limit in sport mode. There is an indicator (where there is the compas) that show the % of power used by the Spark. You probably reached 100% with your brake and rotate moves and the Spark flight controller was no more able to keep it under control resulting in a ground crash. At a point it was upside down and the motors stopped.

Maybe there will be a better flight analyst than me that could check it too but from the flight record, there was no issue with the battery, GPS, RC or the compass.
It seems that the Spark is not pilot proof in Sport mode against very aggressive flight move. We need volunteer to test it

 

[Adrian]

Wow thank you for the analysis, if I understood what you said correctly then this was not user/operator error, but rather the Spark firmware should have prevented this from happening but it currently does not and allows these types of manuvers to shut down the copter in midair.

If that's correct then DJI needs to correct the firmware to prevent this, no?

Thanks,

-Adrian

 

[Dronason]

I am not sure if it really had shut down the motor in the air. It shut down the motors when it was upside down as it should. But I had defined the ground hit as moment the accelerometer was indicating 10g, motors stopped after that. The altitude was negative, probably not flat ground, so cannot clearly use it as reference.
I tried to reproduce the same sticks movements (on grass) but was not able to get same crash. I was also on the firmware release of today, maybe there is now a power limitation to avoid such.
I will retrieve the log file on mine to compare.

 

[Adrian]

I updated the firmware as well today and took for a stress test today, I couldn't replicate the issue. I think they may have fixed it in this firmware as it doesn't seem to tilt/turn as aggressively as it did before.
Fingers crossed that they actually fixed it. Not sure if this is standard for Spark, but I noticed that there was a date that you needed to upgrade by or the Spark won't fly anymore, this leads me to suspect that they fixed something major and crash related.

I will test it some more in the grass to see, but so far so good.

-Adrian

 

[Dronason]

On mine, it never tilted more than ~30% during acceleration and braking, which is the expected value. Also power on battery was up to 80 W versus 100 W in your case. Same for motor, up to 92%, not 100%. So in my case, it stays in a reasonable flight enveloppe where the flight controller keep the control of the Spark.
If there would be a strong wind from the side, it could have maybe changed the game. Did you had high wind during the crash?

 

[wyx087]

May I ask what software are you using to generate those analytical graphs?

 

[Dronason]

May I ask what software are you using to generate those analytical graphs?

You can also do it yourself from the DAT file and plot flight parameters using the CsvView tools available HERE.

DAT files (Spark flight record)
DJI assistant generate a DAT file that can have several flight records in it.
There are 2 type of data in a flight, there is FC (Flight Control) and Vision (Obstacle avoidance) but only FC is used.
The tools will extract the flight record from a DJI assistant DAT file into DAT file like FLYnnn.DAT and will use that file.

There are:
CsvView will plot the first flight record found in the DAT file.
ExtractDJI will split one DAT files containing flight records into one DAT file for each flight.
DatCon is a converter tool to generate various file format.

You must read the documentation on the web site before trying CsvView, it helps a lot to start correctly and use the graphical interface.

You can display several plots at same time and the zoom + cursor is sync over the plots.
There is also GeoPlayer that can plot the flight on map, also cursor is sync with plots.
Finally MagDataPlayer (see documentation for details) that plot the magnetic field vector. Note that it is sync with the zoom of the plots. Usually good to zoom on flight at some altitude before activating it so you can evaluate the calibration of the compass out of influence.

TXT files (Go4 app flight record)
You can upload such flight log from your app here DJI Flight Log Viewer. It gives a good overview of the flight.

 

[actionjack]

Wow thank you for the analysis, if I understood what you said correctly then this was not user/operator error, but rather the Spark firmware should have prevented this from happening but it currently does not and allows these types of manuvers to shut down the copter in midair.

If that's correct then DJI needs to correct the firmware to prevent this, no?

Thanks,

-Adrian

That's the point I was making in my post. You can't give us an aggressive flying aircraft with the flip over shut off feature. If the aircraft, in sport mode can reach the angle needed to shut the motors off mid flight than the 2 should not coexist.

 

[Dronason]

That's the point I was making in my post. You can't give us an aggressive flying aircraft with the flip over shut off feature. If the aircraft, in sport mode can reach the angle needed to shut the motors off mid flight than the 2 should not coexist.

The Spark turn-off at ±90° when just rotated by hand. I am sure that in sport mode, you will be happy a day or another that the motor automatically shut off, for sure not when in the air.
I tried with latest firmware to reproduce aggressive flights but it didn't result in a tilt of more than ±45°. It can be for sure more aggressive and maybe with some extra wind it can influence to the worst. It is supposed to limit the tilt angle to some higher limits than in normal mode w/wo obstacle avoidance, but still being safe.
If you have a Spark not updated at the moment, you can try and repeat it after update to see if there is a difference.
In any case the way the Spark update is a forced one make it clear that there were severe issues with previous version. As there is no way to escape it, best is to focus on the new version and if issues are solved.

 

[actionjack]

The Spark turn-off at ±90° when just rotated by hand. I am sure that in sport mode, you will be happy a day or another that the motor automatically shut off, for sure not when in the air.
I tried with latest firmware to reproduce aggressive flights but it didn't result in a tilt of more than ±45°. It can be for sure more aggressive and maybe with some extra wind it can influence to the worst. It is supposed to limit the tilt angle to some higher limits than in normal mode w/wo obstacle avoidance, but still being safe.
If you have a Spark not updated at the moment, you can try and repeat it after update to see if there is a difference.
In any case the way the Spark update is a forced one make it clear that there were severe issues with previous version. As there is no way to escape it, best is to focus on the new version and if issues are solved.

I have no idea what you said but this thread exists because flying in Sport Mode can create enough angle to turn off the motors. IMHO that shouldn't be. An update to eliminate or appropriately modify that "feature" when in Sport Mode is suggested strongly.