Hi All,

I thought that I would share my story and seek advice.

After having my phantom 3 advanced for 3 weeks I decided to upgrade the firmware this week. I was debating whether or not to do this but after DJI posted the forum thread rationalizing a lot of the complaints that have been posted on the forums I figured that it would be safe.

I took it out for a flight yesterday in an open field. It flew around 40m and then turned sidways and fell around 10m to the ground. No collisions, no CSC, no wind, just flipped on its side and fell to the ground. The gimbal is bent and cracked, shell is cracked, motors make a strange noise, landing gear is bent and wires are popping out of the shell. Basically the sum of all fears for a guy who has saved up for something like this and only had it for 3 weeks.

I purchased the drone from HeliDirect in MA but I'm currently in South Africa and will be here for a long time so I can't take advantage of any local US support. I'm absolutely confident that the crash was due to an internal malfunction of some sort but I'm interested to know what experiences people have had with making a warranty claim to DJI and what data I should try to extract from the drone in order for this to be as efficient as possible. Also if anyone knows of how to make a warranty claim outside of the US that would be helpful too.

Thanks in advance.
Luke.

Edit: Some additional info about the flight.

I always ensure that everything is perfect just before a flight. I'd callibrated the compass (with all of my electronics safely tucked a few meters away), checked that the IMU did not need callibration, ensured that there were plenty of GPS satellites in view and that the signal was strong, hand tightened each of the props, etc...

Interestingly enough the flight logs on my iPad seem to have a bunch of missing data. It's like the phantom jumps from one point to the other on the map just before it flips on its side and dives. Below is a screenshot and I'll see if I can capture the video off my iPad screen.
http://youtu.be/8AkStJZ0rRY <-- Video of the flight log from the pilot app. Watch at the end of the flight. The aircraft seems to just jump from one point to another, flip direction and then it goes down.

Sorry that happened mate, hope you could fix it soon...

leoamartinez@gm Posted at 2015-7-12 22:46
Sorry that happened mate, hope you could fix it soon...

Thanks for the support. I'll post to this thread to give feedback on how the support process works out with DJI. I've seen some good reports so let's hope that mine goes well too

Can you post the log from the controller and P3? Also the cache video from the flight? Should show its final moment.

Sudden drops from the sky for seemingly no reason have happened to some when the battery not being clipped in properly and coming loose... Maybe not the case but falling from the sky is not that common a "fault".
A recent one I also saw was a prop comming off and nosediving the P3..

Dude something similar nearly just happened to me, what a bummer, somewhat shocked here.  Needless to say I am not happy

sploodge Posted at 2015-7-12 23:03
Can you post the log from the controller and P3? Also the cache video from the flight? Should show i ...

How do I retrieve the log from the controller and the P3? I'll google in the meanwhile and see what I can find...

sploodge Posted at 2015-7-12 23:03
Can you post the log from the controller and P3? Also the cache video from the flight? Should show i ...

Also would it cache video if I was not yet recording and if so where would I find the cache? I've only been able to find cache once I hit the record button.

sploodge Posted at 2015-7-12 23:03
Can you post the log from the controller and P3? Also the cache video from the flight? Should show i ...

OK so I found the log from the P3 and I'm about to view it with the maps made easy viewer. I'm 100% sure that it had nothing to do with the battery. I'm always anal about checking that because of stories that I've read on this forum. Interestingly enough it seems as if I'm not the only one that this has happened to recently...check this out...

http://forum.dji.com/thread-23035-1-1.html
EDIT: Dropbox link for log file is https://www.dropbox.com/s/1m9rdfgkbjcgb8c/FLY034.zip?dl=0
When you process the file using maps made easy (http://forum.dji.com/thread-16195-1-1.html) the flight path does not match the path that the Pilot app shows. To me that is indicitave that there must have been some kind of internal error on the drone.

I am sorry to read about your loss.  Unfortunately it sounds like this will be a DJI warranty issue.

Sorry to hear mate. Be interesting to see what your log analysis shows up.

Had you performed a new calibration immediately before this flight, or was this simply another flight at a spot you'd flown previously?

Are you sure that's the DAT file for the flight? The time does not match up to the image you posted initially..

can you get the logs from the ipad. This will show a lot of variables..

But ultimately, contact DJI. They are the only ones that can definitively give you an answer.

sploodge Posted at 2015-7-12 23:46
Are you sure that's the DAT file for the flight? The time does not match up to the image you posted  ...

This is definitely the correct log file. The date/time stamp of the file is the same as that shown within the pilot app. The maps made easy decoder always just starts from 12:00pm.

Check out the altitude. It's GPS so it's a little off but basically you can see it rise and fall within the 13s period that the log on the pilot app shows. Thereafter it sits relatively still except for one spike on the speed graph. That could be me picking it up (with man tears welling) or another malfunction with the drone.

Don't you want fly036.dat...seems you grabbed 034...?

Yeah, upload 35 and 36..

ScottyT Posted at 2015-7-12 23:46
Sorry to hear mate. Be interesting to see what your log analysis shows up.

Had you performed a new  ...

I performed a new compass calibration before this flight. I normally do one before every flight as per the DJI manual just to be safe...sweet irony

sploodge Posted at 2015-7-13 00:06
Yeah, upload 35 and 36..

So the plot thickens...

I forgot to mention that although the drone is badly damaged it still flies. I took it for a flight directly after the crash but the GPS would not lock and it kind of just drifted around about a meter off the ground until I brought it down. This happened at 12:55 as you will see if you look at the log in the video I posted. Interestingly enough though this flight was actually given a time and date stamp that was meant for the flight where the crash happened whereas the crash file (FLY033.DAT) was given the time and date stamp of a flight that happened the day before. It could be that the crash caused this. I'm not sure.

In any event the log seems to catch up with FLY036.DAT which is completely corrupted and unreadable.

I was actually at a wedding so at 1:00pm I went inside for the service. Afterwards I came out to see if I could get the phantom working enough for the shoot and took it on a short flip but the camera feed was too shaky. This is what FLY035.DAT records.  
https://www.dropbox.com/s/vqz50zo5fgatjio/FLY033.zip?dl=0 <-- FLY033.DAT - The crash which was given the time and date stamp for a flight that happened the previous day.https://www.dropbox.com/s/uckc1rbgcra0717/FLY036.zip?dl=0 <-- FLY036.DAT - The file that cannot be decoded. Seems to be the drone catching up.https://www.dropbox.com/s/i5kwm1hvnsgnbab/FLY035.zip?dl=0 <-- FLY035.DAT - The file that shows the flight that I took after the wedding service. Was checking to see if the drone would still be able to record some footage. Unfortunately the gimbal was too damaged

Your lucky it didnt hit the bride dude.  I was meant to go film a breeders horses, thank god I was in my garden or there could have been some serious damage to horses worth millions.  

Try this method instead http://forum.dji.com/thread-22155-1-1.html

You can also convert the dat file to CSV at http://flylitchi.com/logs but it only returns basic information.

From either site, load the resultant CSV in to www.flightreplay.com to get everything plotted.

ScottyT Posted at 2015-7-13 01:36
Try this method instead http://forum.dji.com/thread-22155-1-1.html

You can also convert the dat fil ...

Thanks! Tried it with the 033 file but all it gave was a txt output with an event log. Trying with 034 now.
Same thing with 034, no CSV. Have emailed DJI. Let's see what they say.

looxonline@gmai Posted at 2015-7-13 03:38
Thanks! Tried it with the 033 file but all it gave was a txt output with an event log. Trying with  ...

Yeah I just tried with your file, got "Detected Type: DJI Flight Controller Binary Log". Not sure if it's corrupt or the web site is having issues. Will try with one of mine later.

Another update:

I pulled the .txt from the ipad and compared it to the logs on the aircraft. Seems like the aircraft was out of synch. It has the crash recorded in FLY033.DAT which, according to it, happened on the 10th whereas the iPad has the crash in a .txt file which is stamped with the correct date and time (11th at 12:48pm). The link to the crash is shown.

https://www.flightreplay.com/share/track/actual-crash

Some interesting points:

If you watch the video of the replay with the stick signals at this link (http://youtu.be/8AkStJZ0rRY) you will see that there is never any signal to descend. If you look at the flight data the aircraft makes a rapid descent in the last second or so. Two conclusions can be drawn from this; 1.) The aircraft descended without any command from the stick telling it to do so. 2.) The aircraft still had a fully functional battery making perfect contact since it was able to record the drop until it hit the ground. The end altitude differs from the start altitude because a.) it crashed a few m higher than where it took off b.) GPS is notoriously poor at measuring altitude accurately and I'm guessing that this signal is from the GPS and not the barometer.

So the net out is that the data verifies this as being a 100% bona fide "falling from the sky" event. Interesting that the flight logs from the aircraft and from the app are somehow out of synch. This tells me that something was not right. Also interesting that it happened the very first outdoor flight that I made after updating the firmware...

UPDATE 15/07/2015: I've been speaking to DJI since Monday which makes it only two days since I submitted my query and they have come back to me this morning letting me know that they found that the accelerometer suddenly went faulty during the flight which resulted in the drone dive bombing into the ground.

They have offered to repair it for free. All in all I've been impressed with their response time and service. Kudos to DJI.

looxonline@gmai Posted at 2015-7-15 15:44
UPDATE 15/07/2015: I've been speaking to DJI since Monday which makes it only two days since I submi ...

Well that's good to hear. Bit of a worry that such things can happen - some of these sensors need redundancy. But hey, nobody was hurt.

Do they pay for you to send it back or do you wear that?

How many people saw you crash it?

That's great news.  I bet you feel better confirming what you already knew about the crash, it wasn't your fault.  There aren't too many reports online that show a suspected, and verified product failure.  It's great to know, when that does happen, DJI is making it right.

One thing about this occurrence, I bet you learned more about the internal workings, logs, etc, then you ever expected to learn at this point.  I know I appreciated reading your updates.  Much better for you to crash and me learn vicariously of all the steps of diagnosis necessary, and the conclusions made....then me having to crash and learn those things first hand! <grin>.  Thanks for the updates.

Here's hoping you get your bird back, and up in the air, soon!

I'm sorry for your loss bro and at the same time happy that your bird will be fixed for free. You will still be left without Phantom for the rest of the summer...
Nevertheless, I'm still supprised that some people decide to upgrade the firmware even AFTER reading on forums that it is faulty.
I have mine on 1.1.9 and not willing to upgrade until DJI will come with good firmware.
If it ain't broke, don't fix it (or rather "worsen it" in this case)
Peace.

ScottyT Posted at 2015-7-15 16:23
Well that's good to hear. Bit of a worry that such things can happen - some of these sensors need  ...

Agreed. I work as an engineer and I've worked with accelerometers and gyros in the past and in every instance we have just known never to trust the data coming from a MEMS sensor. Sometimes they just spit out a few rubbish samples and then come right. Best practice is to fit the data to a polynomial using a best fit algorithm or to have redundant sensors.

Unfortunately I have to foot the bill for sending it back. Being in South Africa at the moment and having to send it back to Germany is going to cost quite a lot but it's still less than a new phantom

There were a ton of people that saw the crash. All of them were just as bemused as I was. If you ask me, it was something to do with the new FW. The P3 was flying just fine running 1.1.9

TAZ Posted at 2015-7-15 16:26
That's great news.  I bet you feel better confirming what you already knew about the crash, it wasn' ...

100%. The only thing I never found was a decent decoder for the DAT files that gives access to the sensor data but I may spend some time this weekend trying to decode it. It'll be difficult without a packet structure document but not impossible.

One thing that I never posted was this REALLY useful decoder tool that gives you a whole lot of data from the pilot app txt files (http://thecutter.net/djiconverte ... =%2Fdjiconverter%2F). I've attached the crash log after the decode with some graphs for interest (https://www.dropbox.com/s/2scz9v ... t-Details.xlsx?dl=0). Notice how just before the phantom rolls over the battery reports back as OK and the overall system status is healthy. Then, suddenly the sensor data shows it rotating and the Z velocity starts to increase rapidly...

Vampiszon Posted at 2015-7-15 17:27
I'm sorry for your loss bro and at the same time happy that your bird will be fixed for free. You wi ...

Right you are. I held off and then took the DJI post trying to reassure us as a sign that maybe there were just too many naysayers on the forum...lesson learnt the hard way

Vampiszon Posted at 2015-7-15 17:27
I'm sorry for your loss bro and at the same time happy that your bird will be fixed for free. You wi ...

Indeed. Not sure why they just didn't pull the firmware from the site after they acknowledged there were issues with it. Perhaps they had fixed fatal bugs that they just didn't want to roll back.

looxonline@gmai Posted at 2015-7-15 15:44
UPDATE 15/07/2015: I've been speaking to DJI since Monday which makes it only two days since I submi ...

Sorry to hear about your crash, but it's great that DJI stepped up to make it right for you.
Quick question, though. In your previous post DJI points to a faulty accelerometer as the problem causing the crash. Is this not a hardware failure as opposed to a firmware problem?
Just call me curious.
Hope they expedite the repairs quickly for you!

headkumquat Posted at 2015-7-16 13:14
Sorry to hear about your crash, but it's great that DJI stepped up to make it right for you.
Quick ...

Well it's kind of both. Like I mentioned earlier, accelerometers and other MEMS sensors can sometimes spit out rubbish data for a few samples. This could be seen as a hardware failure but if it is a common issue with the sensor then software should be written to work around it. Given that the aircraft flew after the crash (albeit a bit shaky) it shows that the accelerometer went back to normal.

To me it seems as if the new firmware is a bit more sensitive to bad data coming from the sensors. So it seems more like a blend of hardware/firmware.

looxonline@gmai Posted at 2015-7-15 15:44
UPDATE 15/07/2015: I've been speaking to DJI since Monday which makes it only two days since I submi ...

Good to hear that. Did you speak with DJI after contact them through email? Could you share how you contacted them? There's another faulty crash that needs to be followed up.. http://forum.dji.com/thread-24182-1-1.html

realsy Posted at 2015-7-28 02:02
Good to hear that. Did you speak with DJI after contact them through email? Could you share how yo ...

I spoke to DJI China via email who took 2 days to analyse the data and confirm that it was product failure and then I called DJI Germany who confirmed I could send it to them for repair. It is currently at DJI in Germany and will be there for about 20 working days. I've been assured that the repair will be for free and will update the thread when I hear from them again.

It was actually all quite painless aside from the fact that I had to pay to get the unit there.

looxonline Posted at 2015-7-28 23:29
I spoke to DJI China via email who took 2 days to analyse the data and confirm that it was product ...

I would like to see them simply replace your unit today, not repair a crashed unit, or send you a refurbished unit.  You have done a load of work to support your case.  If I were DJI boss, I'd be expediting a new P3, and adding a couple free batteries for your "pain and suffering".  Not kidding here.  

This is not at all to suggest DJI has not promptly addressed the issue (they did cause it).  I am not suggesting a windfall remedy.  It just seems reasonable you should receive a new in box unit to get you back to where you started, asap.

Let us know when the repaired/replaced unit arrives.

Thank you for sharing your thoroughly documented experience.   

Similar issues with Phantom 3 mysteriously crashing:

looxonline Posted at 2015-7-28 23:29
I spoke to DJI China via email who took 2 days to analyse the data and confirm that it was product ...

Il 31.07.2015 stavo regolarmente volando con il mio phantom 3 advanced quando dopo un minuto è letteramente impazzito e volava da solo senza alcun comando tramite il mio radiocomando...non sono riuscito a controllarlo ed è andato a finire contro un albero cadendo e rompendo l'asse retro del gimbal..ho volato altre volte e non mi dava problemi..solo un'altra volta ho avuto problemi di giroscopio ma nulla di grave..non riesco a capire cosa sia successo...GPS OK, calibrazione OK, no campi geomagnetici...ma dopo un minuto è impazzito...non sono riuscito a farlo atterrare perchè proprio non mi rispondeva, volava da solo facendo giravolte... ho dato 2 volte il go-home ma niente, non mi rispondeva.Ho contattato il supporto dji ma nessuna risposta.. Vi chiedo di aiutarmi.. Allego l'ultimo registro di volo così da capire cosa sia successo..[url][/url]