Mavic just kinda...well fell out of the sky. Here are my logs.

[eEridani]

At some point everything is speculation unless the failure was instrumented and observed.

A propeller that is broken, but still attached and flagging, will impede the motor spin up. It's a simple fact of physics. Once the prop separates, then the motor can spin up. That is what I see the logs telling me. At no time do I see any hint of the motor itself slowing or being part of the problem.

But, that said, before flying again I'd certainly inspect for motor damage. A flagging prop probably didn't do it any good.

 

[slup]

... At no time do I see any hint of the motor itself slowing or being part of the problem.

Then you just speculate about another reason that can "brake" a motor, making it not go to target rpm & with increased current draw ... a reason that in theory & with a good amount of imagination could work, but is against all what we have seen here at the forum so far when it comes to thrust problems, props are rarely "flagging" for 2,5sec in 8000rpm, they break off pretty instantly.

 

[BudWalker]

@ToasteyOstey higher frequency data would be helpful in determining what happened. Said data exists on the Mavic Pro itself and can be retrieved by following the method described here
How to retrieve a V3 .DAT File from the AC
The on board .DAT will be much larger than the mobile device .DAT so you will need to provide it via a public sharing site like DropBox or GoogleDrive. It will be FLY080.DAT.

If the left front prop was missing a blade for the initial 1.9 secs then that might be seen in the higher frequency accelerometer and gyro data. The data may be fast enough to show the accelerometer and/or gyro being noisier during this interval.

 

[slup]

Good idea to settle the said missing data. But on the other hand one missing blade wouldn't make the motor disobey the FC command.

 

[BudWalker]

Good idea to settle the said missing data. But on the other hand one missing blade wouldn't make the motor disobey the FC command.

I wasn't suggesting that the motor isn't following the FC command. It looks to me like in the interval [326.418, 328.26] there is extra drag. After that there is almost no drag.

I'm not too keen on the cause being part of the prop missing but I don't have a better explanation. A missing blade will lower the aerodynamic drag (by half). But, the imbalance will cause energy to be transferred into the airframe - effectively increasing the drag seen by the motor. I'm guessing the latter is much less than the former.

 

[eEridani]

We can't see what happens first, not enough timing resolution. But simple drag on one motor/prop - without a commensurate speed reduction - probably won't cause the drone to yaw, then pitch and roll. And that is exactly what the drone did. So it is physics. Either something hit the drone to rotate it in space, or a prop failed or was forced back into the folded position and/or part of it thrown off and the drone was reacting to that energy. The FC reacted by trying to add power to the affected arm. I'm inclined to this being a bird strike. A prop failure is my second choice. I am disinclined to think anything motor related is involved - there is simply no evidence the motor slowed to require a FC correction.

 

[Sunshynff]

What I thought when @slup mentioned temporary obstruction in his analysis.
Would / could a bird hitting the motor / prop LH front have read like this, or is there too much delay in the timing of the obstruction / recovery and ongoing crash ?
Could the OP see any birds around, near the drone etc, if they were within decent VLOS ?

It wasn't an obstruction, foreign object or bird strike, none of those cause the data to look like that.

 

[Sunshynff]

I am curious with the line L what could have happened when it mentioned it detected a side shock. Would this happen from a prop failing?

I doubt VERY seriously a prop failed if they were stock props, the foldable type, I've had several Mavics, and several other quads with "folding style" props, then you add in DJI's locking system. They get chipped and banged up, sure, but failure, nope.

What did you mean you found a couple busted props, when you retrieved it? Do you mean that the hub was still on the motor, and one or both the blades snapped off? Or was the whole hub/props assembly missing? Cause if you lose a whole prop while flying, it ain't landing anywhere near the mavic. and if one blade from a prop maybe broke off, that would be too light, no way it'd take you're bird down unless it had prior problems.

I've been flying a long time, and had my one mavic almost the whole time, the only incident I've seen props really come apart and ruin someone's weekend, it was my weekend..lol, when I was trying out a set of carbon fiber tri-blades. At that time I had been flying and building freestyle quads for a little while too, and I was doing what you should never do with photog drone, and trying to look cool in front of some buddies, that didn't fly freestyle or any of that, so like a horses back side, I had it in sport mode and buzzing around my truck (empty parking lot) so I could land in the bed or some dumb landing I was trying, and the very edge of one blade struck the very top of a CB antenna I use to keep on roof of my truck. That single blade shattered instantly, sending shrapnel in to the other 2 blades, which sent out more pieces, and it ended up destroying 3 of the 4 props and a perfectly good pair of underwear! Despite us all being well clear of the truck, and landing area, the noise was something I never heard from a drone along with all the shrapnel of pieces, It scared the **** out of me. Mavic was fine, flight wise, couple shards took out two cable ribbons, so I had to replace the those, but had scratch on my lens so just swapped out to a new camera

See what showing off gets ya kids? Nuttin' but looking like a fool and less money in the bank.

 

[Sunshynff]

The prop speed doesn't indicate a stuck motor to me - it is spinning the entire time. The FC increases drive to max - the motor speed increases, albeit slowly, as if there is added drag. A flagging prop would do that.

But the timing seems better than I'd expect for the controller to detect a broken prop and respond. Something isn't making sense in the data provided.

Also to ToasteyOstey and slup

Can I ask you all what software are you using to analyze and the pictures you're posting from the flight log and .DAT files? I promise I'm not trying to "that guy" or sound like I'm talking down to anyone, there's already enough infighting in our hobby among the branches of of drone flying, but have any of you used Betaflight BlackBox? In fact I don't even know if it only reads Beteflight FCs or if you can enter any data in it. I've just never really analyzed my flights with my mavic, because it's usually pretty strait forward flying, so I'm not really familiar with what DJI uses or 3rd party software. With blackbox, which I still don't know half of it's capabilities yet, we use it to analyze very small differences, and patterns, in order to get the smoothest best feeling flights possible. I'm telling ya, there's nothing like finally finding the sweet spot on a new quad, where it's like your hands the radio and quad are all one piece, To put it like a Bostonian, it's a wicked pissah, like flying in creamy buddaah... lol.


I digress though, what you have there is most likely a D-sync or bad motor, or a D-sync from either a bad motor or a bad ESC. Whichever or whatever combo of the three, the flight data is very similar and recognizable after you've seen it a bunch. You all made some good points though, and remember, the graphical image we are seeing, for metric's like throttle percentage/stick deflection, PWM, current are things that the FC is calling for, and I think eEridani mentioned this also, the motor might not be able to deliver, but the FC stills calls for it. Now motor RPM/speed are real time of actual motor, same with current I think, although back in the day those were all interpretations, by most hobby quad pilots, the technology was not there yet, we used to have a set of 3-5 wires from each ESC ran to the flight controller, and that was just to fly and maybe get a Bat voltage and time elapsed clock in your goggles.

Before our drones got high tech and had PID loops we could adjust, if you had a bad motor, with no ability to adjust the PID, or no PIDs at all, that would cause the drone to dip towards the bad motor then fall out of the sky. Today with advanced flying technology, if there's a bad motor, the PID loop will adjust for it by lowering output of the other three motors, and sometimes that is very subtle, and hard to recognize unless you analyze your flight data all the time, or until it fails and your drone drops outta the sky.
A D-sync is caused by the derivative, (D term) being set to high (I think) or either by a few other PID settings Betaflight has that affect D-term down the line. For whatever reason, too high of a D-term in your PID loop, can cause a one motor to fall outta sync, or either for PID loop system not to keep up and one motor falls out sync, The PID loop recognizes this, and has the FC call for more power, hence we see that major spike to 100% in the pwm or call for power, in the graph, for motor number #4 (front left), and why the current and rpm's only slowly followed suite. Then when it "gave up" was the motor not being able to physically complete what the FC and PID loop needed it to do. All this happens in milliseconds, causes the drone to dip towards the bad motor or motor out of sync due to D-term, and almost ALWAYS results in what the FPV guys call a death roll, not a huge deal when those are built to crash and can be fixed fairly easily, not so much on a thousand + dollar investment.

Here's a really good video JB put out a couple of years ago, if you don't know Joshua, he's the go to guy for all things FPV, very smart, but doesn't talk down to people. It's one of his shorter ones, he just covers the basics of BlackBox, and what it looks like to have a D-sync or failed motor, 'bout 8min Long. Joshua Bardwell quick blackbox/D-sync
Oh and this last link is to a short gif or 5sec video that is supposedly a text book D-sync, once lots of factors were examined after several crashes. D-sync gif

 

[slup]

...Can I ask you all what software are you using to analyze...have any of you used Betaflight BlackBox?

Using CsvView to analyze DJI mobile device TXT & DAT + AC DAT's if they aren't encrypted. Get it for free here --> CsvView Downloads

And read up more about DJI logs here --> Mavic Flight Log Retrieval and Analysis Guide

Betaflight Blackbox explorer is used for Betaflight logs... not for logs coming from a DJI photo drone... use it a lot together with PIDtoolbox when I'm tuning my quads.