Mirek6
Second Officer
Flight distance : 609724 ft
Canada
Offline
|
AlexQB Posted at 2018-6-26 14:06
Hello Mirek,
First of all, wooowww !! You left me very impressed! Many thanks for the accurate analysis of the flight, especially for the final moments. What privilege you spent so much time analyzing my fateful flight.
Alex,
Hmmm...
I am no DJI Jedi as BrilhasMuito jokingly said but I used to work as a mountain guide in my previous life. I am not joking here - back in Poland. Beskidy and Tatra. I led groups of university students for several years and took part in mountain guides expedition to High Atlas Mountains in mid-80s. To this day mountains are my passion and I envy you living in such a beautiful spot on Earth.
I do not doubt what you say. I do not doubt max 5 km/hour breeze you felt. However (from my previous life :-) I know how strange and unpredictable weather patterns could be in the mountains. I know and expect different wind speeds at different altitudes (even 10 meters of altitude can make a big difference) and I appreciate wind tunnels created by valleys, crevices, or narrow spaces between rock formations. I know how wind speeds up against rock faces when conditions are right. Spark is light – it does not need big wind to push it around.
But let’s assume that there was really no wind. Not at 18 metres and not below you. How would I explain what I saw? How can Spark in ATTI drift sideways? How can it drift sideways consistent with pitch and roll which I saw at 18 metres altitude while it was standing still which I interpreted as bracing itself against the wind so it does not get blown away?
The only answer could be uncalibrated or damaged IMU. IMU is responsible for keeping the craft level. If it is not level, it will drift in ATTI because the power is distributed to all propellers evenly when there is no input from sticks. So, if, let’s say, the nose of Spark is pointing north and is slightly inclined down, Spark will drift north.
This is less of a problem in P-GPS because Spark does continuous correction of its course based on compass, IMU and GPS. For example, if it starts to drift, it will quickly figure out that there is no input from sticks, GPS position is changing, and it would distribute power to its propellers in such a way that it would drift back. It happens very quickly and constantly.
Is the problem we are seeing here consistent with broken IMU, if there is no wind? Outside of what I saw at 2m 2s – perhaps. But at 2m. 2s when Spark was stationary at 18 metres above you, keeping strong position and staying still while being slightly tilted (a bit of backward pitch and twice as strong roll to the right) can, in my books, be only attributed to bracing against the breeze from NW. Otherwise it would have flow in that direction. So here I must insist that at that point and that altitude Spark was feeling wind.
But forget the (not so strong) wind at 2m 2s. Let’s say no wind below you where Spark had issues.
When Spark’s IMU is unbalanced (not calibrated, has some slight problem) everything may work fine in P-GPS.
Sparks IMU imbalance will be continuously corrected while Spark is flying to keep appropriate position and direction. Until disagreement between compass and IMU is big enough to warrant switching off to ATTI to avoid toilet bowl effect, you are fine.
So, what do we see at 4m 2s? Spark doing almost perfect circle while hunting for stationary position in P-GPS and with no input form sticks. This can happen only if we have a twirl of wind shifting from NE to NW (the direction it is blowing from – and later consistent with direction Spark flew away) or mismatch between compass and IMU starting toilet bowl effect, or (a perfect storm) – both. Knowing mountains (plenty of rocks with iron ore or magnetite) and seeing later compass and yaw errors (in logs), I did bet on wind and/or possible magnetic interference.
But the exact same effect could have been caused by compass / IMU disagreement without the wind (which I noted). If IMU was faulty, and, perhaps, we had additional magnetic interference to completely throw it off – the flight path would be as shown on the picture. After few seconds Spark would have switched to ATTI to avoid progressively worsening situation and progressively wider circles where Spark control would have been very difficult. Its speed was greater than zero around the circle but its GPS position was barely moving – Speed error.
So it did switch to ATTI. Correctly.
If IMU was damaged or uncalibrated causing slight tilt of Spark, it would have drifted. And it did drift in the direction of tilt.
DJI Support will ask you first if you calibrated compass and IMU and if DJI GO App showed them normal when you started your flight. If you did, they showed OK, there was no wind, you have a warranty case of damaged IMU.
And here is a bummer which most likely completely invalidates everything I wrote above regarding IMU malfunction and the tilt (yes, I did keep you in suspense on purpose |
|