I took a 6 month hiatus from flying my Spark but returned to it to take some brilliant fall foliage pictures last week.
No - my drone did not flyaway, as the title may suggest, but I decided to write this post to give a word of advice, again, especially to new drone flyers. This advice may potentially prevent you from unfortunate experience of flyaway.
This forum has seen myriad of articles on drone flyaways, compass, speed and yaw errors. A lot of arguments about how and when to calibrate compass and IMU and what may have caused flyaways experienced by forum members.
I was always a proponent of a "better safe than sorry" practice and a healthy sceptic of DJI GO 4 app accuracy, arguing for compass calibration in certain situations even though DJI GO app showed that all was fine with compass and/or IMU. Such situations included: not using your drone for a few months, travelling long distances (hundreds or thousands of km) to a new location, noticing some weird behaviours during flight etc. Of course, I was also always saying that calibration must be done properly and that bad calibration may cause more damage that no calibration.
But what prompted me to write this post?
Two things: my first flight after 6 month hiatus and an interesting video posted on YouTube few days ago.
When I prepared my Spark for a flight, I checked compass and IMU in DJI GO 4 app - all was fine. Beeing excited about capturing beautiful fall colours, I broke one of my own rules - calibrate after lengthy period of not using your drone. I started to fly and I noticed that my Spark had slight issues in maintaining position. When I was trying to take pictures, it would change position by few metres on its own. I landed, rechecked compass and IMU calibration indicators in DJI GO 4 again - all fine and green. But something was obviously not quite right. I carefully and very slowly, calibrated my compass, right on the spot, in the middle of the woods. Flew again and used all 4 batteries without slightest hint of any issue - solid as a rock.
Than I stumbled upon this video: . Don Joyce talks about his Mavic Pro 2 flyaway at 90km/hour (it ended happily - he did not loose his drone) and his quest to uncover what happened. He manages to explain how a simple yaw error (in his case it may have been a random error and freeze in his Mavic's controller) caused very rapid increase of Mavic speed until it reached maximum velocity without Don touching his controls. He shows detailed graphs and logs and provides rock solid explanation.
Now, his error was unlikely caused by badly calibrated compass or IMU. However, it was a yaw error, which, in great majority of cases, is caused by bad calibration of either compass or IMU or both. He, same as me, advices to do regular calibrations of both compass and IMU either at rugular intervals (he talks about once a month) or calibration at the slightest hint of any trouble in the air.
Again - so I avoid unnecessary arguments here - if you do bad calibration it may hurt more than help. So:
always follow exact steps as described in manual or prompted by DJI GO app calibration process
never do this at home or close to electromagnetic interference or close to any metal structures (in the city you have it all over the place - even sidewalks and curbs may be steel reinforced). Best to do it in the middle of the field somewhere.
don't do it in a terrain which is rich in iron ore, magnetite etc.
keep your phone at least a meter and a half away from Spark being calibrated
make sure you do not have steel or iron jewelerry on yourself during calibration
do the calibration rotation of the Spark very deliberately and very slowly.
JJB* Posted at 10-25 06:07
Nice video, would be better to back up his explantion to see the craft pitch and roll angles....
Yes, it would be.
However, if we believe 100% in what he says (including direction of the wind), and there is no reason why we should not, his explanation really makes sense and roll and pitch values would have to be consistent with this description (we know from which direction the wind was blowing and we know in which direction the drone took off). I can just imagine confusion of the on-board-controller when it thinks that nose of the drone is pointing in the opposite direction than it really is and light wind is pushing the drone in one direction.
There is a very strong positive feedback loop here which goes something like this (from the on-board-controller point of view):
I am drifting to the left without any input from the pilot
Let's roll the AC slightly clockwise so I can get back to where I was
Oopps, I am still drifting to the left even faster now - must be the wind
Let's increase the clockwise roll of the AC so I can get back to where I was
OMG, I am still drfting to the left really fast - the wind must be picking up!
Let's really bank clockwise to hold on for dear life!
With this positive feedback loop which started with on-board controller thinking that the nose of the drone is in different direction than it was in reality (strong yaw error), we have a rapidly escalating flyaway scenario where the drone just takes off with rapidly increasing speed.
Now, in this video, we do not see yaw error indication in the log. AC simply did not know that yaw was bad. So probably a serious on-board controller hiccup rather than compas/IMU calibration issue.
But you can extrapolate this scenario to a compass calibration issue where yaw error shows in logs and, if it is not too serious, drone will start toilet bowl dance or, if it is more serious, as in the video, it will simply rapidly fly away.
In any case - you argued last year for careful compass calibration at the slightest hint of drone stability - I remember your "do it very slowly" instructions (which were spot on!)
I did the same few days ago when I noticed my drone misbehaving while compass and IMU indicators where green. And my issues went away.
This video, in my opinion, really does neatly explain why yaw issue can cause a rapid flyaway.
However, if we believe 100% in what he says (including direction of the wind), and there is no reason why we should not, his explanation really makes sense and roll and pitch values would have to be consistent with this description (we know from which direction the wind was blowing and we know in which direction the drone took off). I can just imagine confusion of the on-board-controller when it thinks that nose of the drone is pointing in the opposite direction than it really is and light wind is pushing the drone in one direction.
Hi Mirek, nice to see you back on here.
I will check his saying with files i have wich where YAW and Gimbal YAW are not the same value...
Speed indication in the flightlog goes funny with errors, when having yaw and IMU errors.
If GPS data is recorded (almost always) this calculated GPS speed differs drom the values in the log.
So his reference to the high increasing speed mayby true but the shows values are perhaps not correct.
And Yes, really sloooooowwwlllyyyy compass swing for my SPARK does do the job better that a quick one!
Hello and good day Mirek6. Thank you for sharing this informative video you have created and for sharing these information with us. Thank you for your support.
Second Officer
Core User of DJI
DJI team