Mavic Air 1 dropped from the sky

[MrRobville]

Firstly, I hope these are the correct logs. I found out in hindsight that the phone I was using had its date and time set back to 2016 for some reason.

Yesterday I was performing a quick flight to inspect the roof of the house. I've done it before and is usually an uneventful procedure.
I fly the drone up one side of the roof to inspect it, then leave it hanging above the nook so that I can walk to the other side of the house and control it from there to view that side.

Only this time it went different. The drone flew fine the first half. Only now, as I was walking towards the other side of the house, I could hear the motors suddenly stop and saw on the video feed that the drone was coming crashing down from the roof. Thankfully it dropped into a flower bed, so it only had some small dents and scratches from the roof tiles as it slid down. After replacing the blades I was able to take it up again without issue.

Another person who was with me was observing the drone on the moment it dropped. He didn't see anything hitting the drone, nor did it make any odd movements- ruling out anything related to a RTH. According to him it simply just plummeted down like a brick without warning after hovering perfectly still.

I'm now doubting if I perhaps accidentally pressed the sticks together to cause a shutdown, but I really can't recall doing that. Also despite being out of sight while walking next to the house, it maintained video feed as I could see the crash before I reached the front of the house.

It's worth noting that the drone hasn't been flown for more than a year. One of the batteries also no longer took a charge. The other two batteries seemed to work fine and are not bloated.

 

[Cafguy]

then leave it hanging above the nook so that I can walk to the other side of the house and control it from there to view that side.

Why not land the Drone while you are changing positions I would say you probably moved a stick OR perhaps a gust of wind traveled over the roof and the Drone got caught in the Downdraft. Good on you for having a spotter. The wind can do funny things on a roof.

 

[MrRobville]

Convenience mostly since it's not a large house and the Mavic holds itself in place very well. When I took it up the second time, I let it hover in front of me so I could give it a good push a few times to see what it would do. Each time it snapped back right where it was. Wind was very mild at that time.

The curious thing is that according to the spotter nothing happened in advance. It really was like as if the power was cut (even though it wasn't since I still had video feed). It wasn't pushed or made any movement before it dropped. Just one moment hovering perfectly still, to falling like a brick the next. The propellers also only had slight scratches from sliding on the roof. No impact damage anywhere. I only replaced them to be safe.

Is it possible to tell from these logs if I've pressed the sticks to stop the motors? If it turns out that I simply made a stupid mistake by not paying attention to my thumbs, then I'd be relieved to know that caused it.

 

[Cafguy]

Post the logs here. Most occurrences like this can be attributed to slight finger movements or "bumping" the stick when you are not aware of it.

 

[MrRobville]

They're attached to the first post

 

[slup]

...I'm now doubting if I perhaps accidentally pressed the sticks together to cause a shutdown, but I really can't recall doing that.

All motors shut off 191,7sec into the flight... but a accidental CSC command wasn't the reason.

According to the DAT log event stream it most probably was a FC failure... likely not a permanent, instead a computational error (as the AC flew normal after the event).

This is from the event stream...



That it looked like a sudden loss of power in the .txt log depends on that this log type only records when the motors are on... the DAT log runs until the AC is powered off.

 

[MrRobville]

Thank you very much for checking out the log. I'm glad to see that it wasn't a user error as I was sure that I didn't touch the controls. I feared the idea of being distracted enough to cause this if I couldn't even remember it. glad that's not the case.

Quite curious to know though what may have caused such a catastrophic failure, since it may very well happen again.
I wonder if the old batteries or the phone acting up had anything to do with it. Or perhaps even the heat of the day since it was a warm day. I recall the motors felt really hot when I grabbed it.

 

[slup]

...Quite curious to know though what may have caused such a catastrophic failure, since it may very well happen again.
I wonder if the old batteries or the phone acting up had anything to do with it...

The log doesn't contain the root cause for the FC failure unfortunately... but this was not battery or phone related at all.

...I'm glad to see that it wasn't a user error as I was sure that I didn't touch the controls...

Well... you actually had a 20% constant negative throttle input going from when you had positioned the drone above the nook until the motors shut off... look at the blue graph below, the chart covers just the last part of the flight & shows all stick inputs, the VPS height &... the barometric height, which shows a slow descent (black graph).

This had nothing to do with the motor shut off though... but it shows that you accidentally pushed on the left stick & that alone could have possibly caused a crash eventually.

(Click on the chart to make it larger)

 

[MrRobville]

Looks like I celebrated too early then. Interesting how the black line contradicts what my spotter recalled. In that case, could it be likely that when it slowly descended, it "landed" on the steeply sloped surface, was forced to tip over because of the continuous input, and therefore shut off all motors as a safety feature? That might explain why there are no impact marks from an "initial drop" out of the air onto the roof tiles. It only has scratches from sliding off.

Isn't that how some people turn off their drones as well, by flipping them over?

Again I really appreciate you sharing this data and looking into it. On a positive note, if it is indeed user error, then I can learn to prevent it next time opposed to some random glitch that would've caused it.

 

[slup]

...could it be likely that when it slowly descended, it "landed" on the steeply sloped surface, was forced to tip over because of the continuous input, and therefore shut off all motors...

Nothing in the log shows that conclusively... but that would have been a likely scenario if there hadn't been a FC failure.

When the errors comes in the DAT log the VPS sensor still reads 1m to what's under the AC (the AC is there directed in the same way as the roof slopes down (meaning longer to the roof for the front side than for the rear side of the AC)... despite this, the right front motor gets the first disturbance through a sudden FC command of 100%, the current draw rises way higher than for the other motors, but the RPM goes instead down slightly... this results in a negative pitch change of 56 degrees (tilts front down) during 0,04sec until the rest of the motors get a command to let of to counter that uncommanded negative pitch, at the same time the AC rolls to the right side

So a reason for the "tilt control failure error" could have been that the right front couldn't rotate freely (the current draw increases)... but that's odd as it was more air under the front than under the rear depending on how the AC was directed in relation to how the roof was tilted, so more likely that the rear would have hit the roof first.

The drone was past the nook & directed according to the red arrow... & tumbled down after the motor shut off towards the street.


Have you checked the right front motor... does it rotate freely, no debris inside? Anything else on that roof besides the roof itself that the right front could have been blocked by?

 

[MrRobville]

When I twist the motors by hand, the front right motor does in fact feel ever so slightly different than the others. The "steps" it makes because of the magnet feel more pronounced in this motor compared to the other three, which feel a little smoother in comparison. No strange sounds though.

The roof itself has fairly recently been replaced and has no debris, dirt or obstructions. When I left the drone hovering to walk to the front, it looked like it had a little more than one meter of clearance with the chimney, which is the only thing that really sticks out.

Could it be that the hot motor that I touched was the front right one? I brushed it off thinking that it was because of the sunny day that the metal felt hot, but perhaps there was something else going on.

I did notice during my second attempt that when the propellers started turning, they made a slightly more noticeable sound as they spun up. Like a slight clacking.