Crashed - but why?

[Frankfurter]

Yesterday I did a first instruction flight in my garden with a beginner, without any problems. After takeoff and the obligatory function check, I switched to tripod mode and it continued to fly.
After changing the battery, I wanted to start again and do the function check.
From that moment everything went wrong: the drone drifted and stopped responding to commands. Result: drone hits tree, crash.
The flight log shows an engine problem and an anomaly in a battery cell.
But I don't understand the real reason.
Any of the cracks here have an idea?

 

[Frankfurter]

Sorry, accidentally I didn't attach the log-files

 

[Yorkshire_Pud]

ooo definitely something weird with the short flight, here is the log on Phantomhelp 9VF4VK98T2UCT9TZFZIQ just look at the heights reported, I assume they are impossible.

second flight DJI Flight Log Viewer - PhantomHelp.com

 

[Frankfurter]

ooo definitely something weird with the short flight, here is the log on Phantomhelp 9VF4VK98T2UCT9TZFZIQ just look at the heights reported, I assume they are impossible.

second flight DJI Flight Log Viewer - PhantomHelp.com

I agree, as far as I remember, I bought a Drone not a Submarine ....
As a matter of fact, this negative altitude was one of the weird things. Another one was, that I tried to "emergency-stop" the props after approx. 5 seconds for the first time, but the commands do not appear in the log. How can that be?

 

[Yorkshire_Pud]

Ah the CSC thing I can probably explain.
With the CSC or mid-air-motor-stop option set to the default of "emergency only" or whatever the term used is, the drone will ONLY stop the motors in mid air if the drone thinks it has suffered an emergency.
To get the response to the CSC to be an actual mid air motor stop and ENTIRELY at your command, you must switch the option to "Anytime" or whatever the term used is......BUT I suspect you would find that the delay period is FAR too short for that option to be set on a permanent basis and probably under 2 seconds.
Between here and the DJI forum that has been discovered to have caused the loss of drones. Hopefully @slup or @Meta4 or @sar104 etc. will be along to look more closely at your logs.

 

[Frankfurter]

Good morning Yorkshire Pud,
thanks for the hint. My point is that I wanted to initiate a CSC immediately when the bird began to behave weird. What I don't understand is the fact, that the rudder inputs recorded in the flight log are completely different from what I actually did and that's quite disturbing ...
I would really appreciate if one of the cracks would have closer look at the logs. Thanks in advance ;-)

 

[Yorkshire_Pud]

I see what you mean. Just for the record where do you think the sticks should be to in the CSC position?

 

[Frankfurter]

Just for the record: both down and inward.
Back to serious. In former times I used to fly gliders and single engine aircraft. One of the things that are anchored deep in my mind since then is to train emergency procedures frequently …
And that’s the irritating point in the story: I know exactly what I did but the logs tell another story.
As far as I can see, there are only three solutions:
1. I can’t trust me and my training any longer or
2. I can’t trust the logs or
3. the RC didn’t work properly.

So my wish is to find out, which one is the right one.

 

[Yorkshire_Pud]

Since you question the accuracy of the logs etc. I would be inclined to test things.

Instead of using the automated motor start, use the CSC to start the motors and hold that CSC for several seconds after the motors have started then release the sticks and then give a second CSC to stop them. Repeat several times.
Look at the logs via uploading them to Phantom help's log viewer and check whether the CSC's are shown.

I use Phantomhelp's log viewer to decrypt my Mini 2 flight logs and it is my recollection that they catch the motor start CSC. I suspect they would also catch the motor stop CSC but If they don't I am fairly sure they will catch some of the movement of the sticks towards the motor stop CSC.

I do not normally use the CSC to stop the motors when the drone is on the ground, rather I just hold the throttle closed so I can not check 'how much' of a motor stopping CSC the logs capture in those circumstances.

Additionally, if you are feeling brave and have a good grip, hand hold the Mavic Air, start the motors and, retaining a good grip of the drone, give it some throttle, enough to have it try to take off.
After it has fought you for a second to two, give it a CSC. If you move your hand during that attempted CSC I suspect the drone will continue to fight you. Release the sticks, hold you hand still and then close the throttle and hold it closed until the motors stop.
Check the log for that test.
I did the latter with a Mavic 2 Zoom whilst investigating the CSC response thing.
With the M2Z's response to a CSC set to "Breakdown"/"Emergency Only" and my hand moving, the CSC would not stop the motors. With the response to a CSC set to "Anytime" or whatever the word, the drone stopped it motors after the CSC had been held for, from memory, around 2 seconds.

If you try the hand holding thing DO NOT leave the drone fighting you for a prolonged period, you may get the ESC's hot.


You could also calibrate the controller, I assume the Mavic Air's Go4 has some means of depicting the movement of the sticks whilst you a 'playing' with them.
If so the calibration will do two things
1) show you if the controller currently indicates the correct response to stick position and movement and
2) possibly corrects that response it is currently in error.

With regards to your 'solutions' 1 to 3.

Regarding the training bit of 1) WERE you aware of the consequence of the CSC if the CSC response setting is "Breakdown"/"Emergency Only"?
If not there is no shame in saying you were not, many people do not know about it.
Besides which there are TWO CSC positions: a) both down and inward & b) both down and outward.
I assume that you fly with the CSC response set to "Breakdown"/"Emergency Only". I assume the Mavic Air behaves in the same way as the M2Z i.e. the 'Anytime' 'delay' is very short and that, if you knew about this, then you either are a VERY careful flyer or consider "anytime" to risky to use all the time.

Regarding the memory bit of 1) we all make mistakes.


Regarding 2) I suppose it is possible for the logs to be incorrect but if so you have happened upon the "one in a million" occurrence of that, hence my suggesting the testing. However see *

Regarding 3) A possibility, hence the checking of the logs of check flights and the suggested calibration.

*From the reported heights I suspect there was at least a barometer glitch, that glitch? may have caused other problems or have been a consequence of other problems but I am not skilled enough to pick such information out of the logs.
@slup, @Meta4, @sar104 etc. probably are skilled enough, hence my trying to attract them to this thread.

 

[Frankfurter]

First of all, thank you very much for your time and effort in replying! My answer is a little late, I had a painful encounter with a wasp.
Back to the subject. The suggestion for the tests makes a lot of sense to me and I will definitely carry them out. After that I should see things more clearly. I also agree with you that we all make mistakes and therefore "solution" 1 is the most likely.
But first I have to have the bird repaired, the rear left motor is obviously defective and the entire arm is no longer properly adjusted.
As soon as the bird flies again I will do the tests and share the result.
As for the various settings for CSC, you got me curious and I re-read the entire DJI Go 4 app manual and the Mavic Air manual. At least in the German language version there is nothing about it :-(

Last but not least, the question of what can have led to these absurd altitude data on the last flight with the crash? I'm stumped and still hoping that SAR104, Meta4 or slup will have a quick look at the issue.