Mavic Pro lost at sea - stalled & dropped into ocean: AirData criteria normal

[wa1pgb]

Hello all.

I am seeking advice and expert evaluation and diagnosis about an incident that transpired on August 1st, 2019 when my Mavic Pro dropped into the ocean unexpectedly and without any apparent reason according to the many evaluation parameters available on AirData.com .

On August 1st, 2019 my Mavic Pro stalled inexplicably and dropped into the ocean off the coast of the Big Island of Hawaii. The Mavic simply stopped and fell 16 feet, it was right in front of me, around 15 feet from shore, directly below where I was standing on a 35 foot high embankment, with 34% battery still remaining. I had just completed two short Litchi missions and a RTH with this battery, and four missions with another battery, and was, at the time of the stall, free flying slowly to capture closeups. I had been carefully controlling the Mavic’s position while continually watching the drone, and when I momentarily glanced at the RC iPad mini 4 screen, I heard a splash, thinking a small rock had fallen, and the drone was gone. I actually saw the splash before I realized that it was the Mavic that had caused it. A few seconds later I received the lost connection warning, but all Litchi other program notifications and "hints" never seemed to recognize that anything at all had happened! (I tried to find the Mavic several times over the next few days, but the waters in this location are treacherous, and the hike down the cliff was perilous as well.)

A sudden drop of this Mavic Pro from a low height had happened to me once before, but that 1st time (September 30, 2018) it was hovering 8 feet over a wooden deck, and with a different battery. After it fell that previous time, when checking the battery cells history on AirData, I found that the battery cell #2 had exhibited continued minor and major deviations, the entire column for this cell was in red, and 19 seconds before the Mavic fell, the cell had dropped down with major deviations below 3vdc, and down to 2.026v. (See screen shots below.) I retired this battery (AirData battery # 2) immediately.

In the present case (August 1, 2019), however, there was no sign in AirData of any problem with the battery, and none whatsoever with any flight, control, or aircraft parameter, criterion or characteristic . Everything was normal and in the “green;” radio signals, GPS, sensors, Voltage/Amperage ratios, battery status - all appeared fine. The Litchi program was still running on the device several hours later, with the screen frozen on the last camera view, after notifying me that I had lost contact. Unfortunately in this case, to be cautious, I was not caching the video. All data that I know how to interpret seems to me to be as if there was no problem with this flight. Everything reads so unnaturally normal as if nothing at all went wrong..

I have included some screen shots below that capture, first, the AirData history for the present incident - August 1st, 2019, and then the AirData history for the earlier September 2019 event. I am happy, of course, to offer anyone to have access to all flight logs, Litchi .csv flight logs, and any Airdata and Dji flight .txt or .dat files. I do not know what specific logs would be most helpful to aid in such a diagnosis.

Please let me know if access to my AirData account would be helpful, and if so how I arrange that. And, please let me know any other logs on my Mac computer would be helpful and in what format.

Many thanks to anyone who might help me understand what went wrong.

I am very grateful.

Lorin Hollander

... Since writing the above post I viewed the flight record .cvs in the 'Phantom Help' online flight analysis website and observed that the VPS altitude readings were almost always very different than the IMU altitude, a problem that has been frequently observed when flying over water. The VPS, in this case, was virtually always measured as 0 feet, perhaps indicating some malfunction. Therefore the problem may not be the state of the battery at all.

I include the Phantom Help link here that can be viewed by others.
DJI Flight Log Viewer - PhantomHelp.com


Screen shots below are:

1) - Present incident battery (AirData Battery #3) - August 1. 2019
a) - AirData Battery # 3 "All Good"
b) - AirData Battery # 3 "Minor deviations"
c) - AirData Battery # 3 "Battery graph"
d) - AirData Battery # 3 "Cell Deviation"
e) - AirData Battery # 3 "Battery Level 34%"

2) - Previous incident defective battery (AirData Battery #2) - September 30, 2018
a) - AirData Battery #2 "Major deviations"
b) - AirData Battery # 2 "Cell Deviation Graph"
c) - AirData Battery # 2 "Battery Deviation"

Please do not confuse the most recent flight data included below, August 1st, 2019, where the Mavic Pro fell into the sea, with the previous aircraft stalling incident, September 30, 2018, where the aircraft simply fell to the ground..

3 ) - 2019-08-01_12-07-55_FLY310.DAT Litchi Flightlog file (This may not be the correct file needed.)
The other Litchi Flight Log is in .csv format which does not seem to be able to be uploaded here.

Please let me know what other files I should upload to this forum to present the information needed to diagnose what happened in this fatal crash.

I am very grateful to all.

Lorin Hollander

 

[Meta4]

2) - Previous incident defective battery (AirData Battery #2) - September 30, 2018
a) - AirData Battery #2 "Major deviations"
b) - AirData Battery # 2 "Cell Deviation Graph"
c) - AirData Battery # 2 "Battery Deviation"

Please do not confuse the most recent flight data included below, August 1st, 2019, where the Mavic Pro fell into the sea, with the previous aircraft stalling incident, September 30, 2018, where the aircraft simply fell to the ground..

Umm ... why post confusing data from an unrelated incident?

The other Litchi Flight Log is in .csv format which does not seem to be able to be uploaded here.

That would be the most useful data.
If you can't post it, put in Dropbox or similar and post a link.

 

[wa1pgb]

Hi Meta4,
Thanks for your reply.

As I wrote in the post, the table of battery voltages that you reference indicating that battery # 2 cell went below 3 volts is the table of the previous battery incident, and not the incident where the drone was lost at sea. That was battery # 3 - which measured no major deviations. I cautioned not to confuse the two incidences I included of AirData information. I included the September 30th data to demonstrate the difference. So, in the present case, the battery voltage was not the problem.

Since writing the above post I viewed the flight record .cvs in the 'Phantom Help' online flight analysis website and observed that the VPS altitude readings were almost always very different than the IMU altitude, a problem that has been observed when flying over water. The VPS was virtually always measured as 0 feet, perhaps indicating some malfunction.

I include the Phantom Help link below that can be viewed by others.

Many thanks, Meta4, I greatly appreciate your tine and interest.

Lorin

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

 

[wa1pgb]

Hi,

To whomever it was that commented on the content of this post, thank you for your input. You wrote:

" ... Umm ... why post confusing data from an unrelated incident?"

Good question, thanks. For me, the two incidents seem closely related in the nearly identical behavior of the Mavic Pro - the engines stalling and the aircraft dropping for reasons that to me are unclear. I thought that including the related nearly identical situation would serve to help in the diagnosis. I of course may be mistaken.

I included the AirData battery information for both incidents to bring clarity to the stark difference between the performance of the obviously compromised battery in the earlier incident and the apparent lack of any such measured battery problem in the recent incident which ended with the loss of the aircraft.

I hope this will not hinder an objective evaluation of the data.

I will post the .csv file via Dropbox as soon as possible. If there is another way to post a .csv file , please let me know

Thanks for your input.

Lorin

 

[wa1pgb]

Umm ... why post confusing data from an unrelated incident?

That would be the most useful data.
If you can't post it, put in Dropbox or similar and post a link.

Hi again,

Here is the link to the .csv data uploaded to Phantom Help. and included in this thread earlier as above.

DJI Flight Log Viewer - PhantomHelp.com

 

[BudWalker]

Hello all.

I am seeking advice and expert evaluation and diagnosis about an incident that transpired on August 1st, 2019 when my Mavic Pro dropped into the ocean unexpectedly and without any apparent reason according to the many evaluation parameters available on AirData.com .

On August 1st, 2019 my Mavic Pro stalled inexplicably and dropped into the ocean off the coast of the Big Island of Hawaii. The Mavic simply stopped and fell 16 feet, it was right in front of me, around 15 feet from shore, directly below where I was standing on a 35 foot high embankment, with 34% battery still remaining. I had just completed two short Litchi missions and a RTH with this battery, and four missions with another battery, and was, at the time of the stall, free flying slowly to capture closeups. I had been carefully controlling the Mavic’s position while continually watching the drone, and when I momentarily glanced at the RC iPad mini 4 screen, I heard a splash, thinking a small rock had fallen, and the drone was gone. I actually saw the splash before I realized that it was the Mavic that had caused it. A few seconds later I received the lost connection warning, but all Litchi other program notifications and "hints" never seemed to recognize that anything at all had happened! (I tried to find the Mavic several times over the next few days, but the waters in this location are treacherous, and the hike down the cliff was perilous as well.)

A sudden drop of this Mavic Pro from a low height had happened to me once before, but that 1st time (September 30, 2018) it was hovering 8 feet over a wooden deck, and with a different battery. After it fell that previous time, when checking the battery cells history on AirData, I found that the battery cell #2 had exhibited continued minor and major deviations, the entire column for this cell was in red, and 19 seconds before the Mavic fell, the cell had dropped down with major deviations below 3vdc, and down to 2.026v. (See screen shots below.) I retired this battery (AirData battery # 2) immediately.

In the present case (August 1, 2019), however, there was no sign in AirData of any problem with the battery, and none whatsoever with any flight, control, or aircraft parameter, criterion or characteristic . Everything was normal and in the “green;” radio signals, GPS, sensors, Voltage/Amperage ratios, battery status - all appeared fine. The Litchi program was still running on the device several hours later, with the screen frozen on the last camera view, after notifying me that I had lost contact. Unfortunately in this case, to be cautious, I was not caching the video. All data that I know how to interpret seems to me to be as if there was no problem with this flight. Everything reads so unnaturally normal as if nothing at all went wrong..

I have included some screen shots below that capture, first, the AirData history for the present incident - August 1st, 2019, and then the AirData history for the earlier September 2019 event. I am happy, of course, to offer anyone to have access to all flight logs, Litchi .csv flight logs, and any Airdata and Dji flight .txt or .dat files. I do not know what specific logs would be most helpful to aid in such a diagnosis.

Please let me know if access to my AirData account would be helpful, and if so how I arrange that. And, please let me know any other logs on my Mac computer would be helpful and in what format.

Many thanks to anyone who might help me understand what went wrong.

I am very grateful.

Lorin Hollander

... Since writing the above post I viewed the flight record .cvs in the 'Phantom Help' online flight analysis website and observed that the VPS altitude readings were almost always very different than the IMU altitude, a problem that has been frequently observed when flying over water. The VPS, in this case, was virtually always measured as 0 feet, perhaps indicating some malfunction. Therefore the problem may not be the state of the battery at all.

I include the Phantom Help link here that can be viewed by others.
DJI Flight Log Viewer - PhantomHelp.com


Screen shots below are:

1) - Present incident battery (AirData Battery #3) - August 1. 2019
a) - AirData Battery # 3 "All Good"
b) - AirData Battery # 3 "Minor deviations"
c) - AirData Battery # 3 "Battery graph"
d) - AirData Battery # 3 "Cell Deviation"
e) - AirData Battery # 3 "Battery Level 34%"

2) - Previous incident defective battery (AirData Battery #2) - September 30, 2018
a) - AirData Battery #2 "Major deviations"
b) - AirData Battery # 2 "Cell Deviation Graph"
c) - AirData Battery # 2 "Battery Deviation"

Please do not confuse the most recent flight data included below, August 1st, 2019, where the Mavic Pro fell into the sea, with the previous aircraft stalling incident, September 30, 2018, where the aircraft simply fell to the ground..

3 ) - 2019-08-01_12-07-55_FLY310.DAT Litchi Flightlog file (This may not be the correct file needed.)
The other Litchi Flight Log is in .csv format which does not seem to be able to be uploaded here.

Please let me know what other files I should upload to this forum to present the information needed to diagnose what happened in this fatal crash.

I am very grateful to all.

Lorin Hollander

The Litchi log files are woefully inadequate for the purpose of analyzing incidents.

I looked at 2019-08-01_12-07-55_FLY310.DAT . That was produced by the DJI Go App (not Litchi) 2019-08-01 22:07:55 GMT. There was no motor start. Here is where it occurred.

 

[sar104]

I don't see any obvious causes in the flight log. The battery cells looked fine. These events usually result from one of two problems - either the battery is not properly installed and dislodges (generally preceded by a sudden flight maneuver that didn't occur here) or the FC crashes (sometimes with precursor indications in the DAT event stream, but Litchi doesn't write one of those). The onboard DAT file would very likely contain the explanation, but without the aircraft that's off the menu.

 

[sar104]

The Litchi log files are woefully inadequate for the purpose of analyzing incidents.

I looked at 2019-08-01_12-07-55_FLY310.DAT . That was produced by the DJI Go App (not Litchi) 2019-08-01 22:07:55 GMT. There was no motor start. Here is where it occurred.
View attachment 79899

It was just a compass calibration.

 

[wa1pgb]

Hi Bud,

Thank you for your input and for your time.

Please explain why you find it was a compass calibration issue. Is that obvious from the files I submitted?

I include here a .pdf of the .csv file, (compressed in Zip) at first glance the font appears tiny and unreadable but when zoomed in it becomes quite clear. I keep noticing what to me are unexpected and incompatible readings in the ever changing differences between the "altitude(feet") and "ultrasonicHeight" columns (columns 3 and 4 from left) which crop up throughout the file. For instance, as in the following screenshots. The final example is from the very end of the flight as the Mavic dropped into the ocean.

Many thanks again Bud. and to all others who have offered ideas and suggestions.


Lorin

 

[BudWalker]

Hi Bud,

Thank you for your input and for your time.

Please explain why you find it was a compass calibration issue. Is that obvious from the files I submitted?

I include here a .pdf of the .csv file, (compressed in Zip) at first glance the font appears tiny and unreadable but when zoomed in it becomes quite clear. I keep noticing what to me are unexpected and incompatible readings in the ever changing differences between the "altitude(feet") and "ultrasonicHeight" columns (columns 3 and 4 from left) which crop up throughout the file. For instance, as in the following screenshots. The final example is from the very end of the flight as the Mavic dropped into the ocean.

Many thanks again Bud. and to all others who have offered ideas and suggestions.


Lorin

I didn't find that it was a compass calibration issue. Maybe you're confusing @sar104 's observation that the .DAT log file you submitted recorded a compass calibration that was performed.

I mentioned that Litchi log files are limited for the purpose of analyzing an incidents. In this particular case I don't think there is enough info to determine what happened.

The ultrasonic sensor is limited in the distances it can measure. It varies, but if the height is above 30 meters or so there isn't enough of a return for the sensor to determine the distance. In this case the ultrasonic height is invalid. Litchi assigns a value of 0.0 to convey that the data is invalid. For your flight the agreement between the altitude and the ultrasonic height was pretty good when the ultrasonic data was valid.

 

[wa1pgb]

Hi Bud,

Thank you again for your reply and for your evaluations and assessments. It is unfortunate, of course, that the information needed to fully determine what ultimately went wrong is not available. It is particularly concerning since the final flight appeared to be extremely stable and without any advanced indication of stress, abnormality or difficulty.

If there is any other information, existing files or records, that I can can make available, please let me know.

Thanks also to sar104.

I am very grateful

Lorin

 

[worldtraveller]

Another case of a DJI drone suddenly stopping flying and falling into the water. This case seems to be similar to other cases on the forums....not an isolated event. I guess that DJI will give their standard reply and will refuse to replace the drone.

 

[gnirtS]

Another case of a DJI drone suddenly stopping flying and falling into the water. This case seems to be similar to other cases on the forums....not an isolated event. I guess that DJI will give their standard reply and will refuse to replace the drone.

Except DJI have replaced drones that have dropped due to no user error. Many of them.
What they wont do is replace drones that are user error or cant be shown not to be.
The main issue with the Litchi log files dont provide enough detail to show if it is or isnt.

A big problem is most people CLAIM no user error (or wont admit it) but on analysis it turns out it was. This does NOT appear to be the case here though.

 

[wa1pgb]

Thank you for your posts.

I did not notify Dji about this incident. Is this something I should do, and if so is there a recognized best way of accomplishing it? Thanks for any information about contacting Dji.

A year ago I attempted several times to receive assistance from Dji regarding a different incident, and a Dji tech told me they would send an email with all instructions, and when checking with them a week later after the email did not arrive, I was told there was no record on file of Dji ever having been contacted about this. I never received the email. At that time my Mavic Pro was still in the Dji insurance/Restore program, at the present time the lost Mavic was no longer insured by Dji.

Thanks again.

Lorin

 

[Snappyman123]

Re DJI replacing drones, I stupidly flew my Phantom 3 Pro into a tree above a canal and it finished up in the canal. I explained what I had done to DJI and they replaced the Drone with A brand new one. Can't fault them.

 

[wa1pgb]

Thank you for your post. This is very encouraging. I will notify Dji this coming week. I had recently a very good experience with Dji when applying the Refresh insurance onto my new Mavic Pro.

Thanks again.

Lorin

 

[gnirtS]

5 more from DJI forum. July and August 18...

 

[Jackcutrone]

5 more what from DJI? I am trying to get a replacement without the old drone. THanks.