Hi there. On march 15th 2017 I did a trip to Merida State, here in  Venezuela. Two days before the trip I did IMU and Gimbal calibrations.  Before the very first flight I did, as usual, the compass calibration. I  took off near a lagoon (Mucubaji) I wanted to film and photograph. I  hover the drone  for about 15 sec to see stability and it was good as always had been. I  started to fly over the lagoon to get a good position before starting  video recording. At about 1 km away from the home  point I had gone to the limit of 120 mt high but I needed to go  higher in order to get a better position for filming. Just after setting   the new limit to 500 mts I started to get weird messages of compass  error and GPS failures; despite my intends to go up it did not respond.  The drone was located on at 120 mt over a border road and at 140 mt  away from power lines (running parallel to the road). At the same time I  started to get an screen message of ATI-Mode. I pressed the RTH button  and nothing happened also the drone was not responding to the stick  commands. All happened very fast and I started to see weird  gimbal ups and downs, and, in a second the drone started to roll over  and  over until I lost the signal. At that moment I knew I has lost it due to  a crash. It was aprox at 1.5 Km away and after an hour we were able to  recover it at aprox 20 mts from the power lines with the battery out. I  managed to get the  battery inserted again and it did start with normal sounds,  but, it did not communicate with the RC. After returning to Caracas, my  home town, I managed to open the P3P to see it there was any  disconnected cable. To my despair I found the two main motherboard  connectors loose (broken) due to the impact at more than 100 mts high.  At that moment I came across that my P3P was completely out of service  with not feasible way to repair it myself. I have analyzed the track log  and  found the drone had flown away at max speed before crashing on the  ground. The flight log is available in my DJI account. The P3P is out of   warranty now, I wonder if this kind of malfunctioning is covered is some  way by DJI. I  contacted support asking for them to analyze the log in order to have an  answer about that. Support answer they need to have the drone in the US  repair service just to check out the log. I would like to have an idea  how much it is going to cost before deciding it it is worth spending  the repair cost (plus sending tu the US and getting back to Venezuela). It has broken hull,  camera broken at the gimbal base and broken main board. I know DJI can check flight logs in  advance, I just saw it in another thread. By the way, I bought a Phantom  4 Pro to continue my work as a Pro Photographer, nevertheless, I got it  2 days ago and in the very first flight it crashed from 2 mts height  hovering as a pre-flight procedure. I posted it here: https://forum.dji.com/thread-94481-1-1.html. I am frustrated and very disappointed about my experience with DJI drones so far.

I just came across, thanks to Hallmark007 that I can share the flight log of this last flight, here it is:

https://healthydrones.com/main?flight=ff03538d9581777cf1102a723d560278

Any comments or evaluation will be welcome.

We feel very sorry for your crash.
Since the drone is out of warranty, we will not provide the data analysis service, which you can get it on DJI official website: http://www.dji.com/service/policy
To ensure the safety and performance, we do suggest you send it in, our engineers will evaluate it firstly then send you an e-mail for the invoice.
You can submit a case number online: https://repair.dji.com/en/SelfRepair/Area

I have checked your flight log for you.  It accurately records your flight and it shows the flight exactly as you described.  Thanks for a very detailed description of your observations for a change!

Everything ran perfectly up until you changed the altitude setting.  During this time your speed slowed to 0 mph while holding an altitude of 390ft.  It continued correctly for another 37 seconds at which point you applied a little left stick up for a short period.   There were no further stick inputs recorded for the remainder of the flight.   At this point we see the aircraft rotating rapidly and the camera gimbal go out of the normal operating range of -30° to 90°.  This is a typical death-spiral as the aircraft goes out of control prior to a crash, and the aircraft tumbles and inverts.  However, incredibly, your aircraft quickly recovered from this and actually gained 15ft altitude!  At this point, I thought you may have had a small birdstrike, sufficient to upset the flight, but minor enough to be able to quickly recover.  Your speed varied betweed 0-20mph as it retained an altitude of 400ft.

At 306 seconds into the flight, it records your RTH "go_home" command.  It turns to point approximately towards home, facing around 110°, while still drifting roughly West.  At 316 seconds we see yet another spin and tumble, this time it fell in altitude to 95 feet, impacting with the ground and ascending again to 172.5ft before the record stops suddenly, showing tumbling and rotating the whole time.  It may be possible there was a second birdstrike which took the aircraft to the ground, damaging it, and it then tried to take off again to RTH, but the battery eventually fell out.  The reasoning is complete speculation by me to try and fit the data.  I have no evidence of a birdstrike.  At 95ft, the kml data (the flightpath) intersects with the surface of the mountain for a short time, indicating it had hit the ground or foliage on the side of the mountain.  Interestingly it was still able to fly and it did ascend, but travelling West, having left RTH mode and entering ATTI.


Another possibility for the loss of control would be severe updrafts in close proximity to the side of the mountain, sufficient to throw the aircraft around and stronger than the 10m/s speed available from the aircraft to attempt to remain stable and come home.  The other thought I had was that the mountain may have large iron ore deposits and had a severe magnetic affect on the aircraft compass causing a loss of orientation and crash.  However I would expect to see many more error messages in that situation.  None of this adequately explains why it had a tendency to drift West, unless there were strong air currents taking it in that direction.

There was a bug in earlier versions of the DJI go software that may fit with some of your circumstances.  I am not sure if this was identified properly and published anywhere, although I believe newer software was quietly released which fixed the issue.  In order to see if this may be the case, can you reply with what device you are using?   Is it Apple iOS or Android?  Can you check what version of the DJI go app you have installed?   What firmware do you have loaded in the aircraft?

As I remember the bug, there were a set of circumstances under Android with an older Go app, and if the aircraft was flown outside of a limit that was set in the app.  The aircraft would create a false home point roughly west 1-2km away.  It's possible changing the limit while in flight upset something.  There is no evidence of that here, but I'm just being thorough and want to rule the possibility out.  Are you certain that you were changing the altitude limit in the app, and not the distance limit, usually associated with limiting the distance flown whilst training?

I'm sorry about your crash, this is indeed a very strange flight and incident overall.  The data is very consistent with very few gaps or error messages indicating breaks in the telemetry.  The fact that you report seeing it tumble in the video suggests you were well within range and had a strong radio signal.  If you can supply the previous information I requested I'll look into this further for you.

In relation to fixing the damage and costs, I'll leave that to others.  Perhaps upload a couple of photos of the damage to give us an idea.

endotherm Posted at 2017-4-30 10:34
I have checked your flight log for you.  It accurately records your flight and it shows the flight exactly as you described.  Thanks for a very detailed description of your observations for a change!

Everything ran perfectly up until you changed the altitude setting.  During this time your speed slowed to 0 mph while holding an altitude of 390ft.  It continued correctly for another 37 seconds at which point you applied a little left stick up for a short period.   There were no further stick inputs recorded for the remainder of the flight.   At this point we see the aircraft rotating rapidly and the camera gimbal go out of the normal operating range of -30° to 90°.  This is a typical death-spiral as the aircraft goes out of control prior to a crash, and the aircraft tumbles and inverts.  However, incredibly, your aircraft quickly recovered from this and actually gained 15ft altitude!  At this point, I thought you may have had a small birdstrike, sufficient to upset the flight, but minor enough to be able to quickly recover.  Your speed varied betweed 0-20mph as it retained an altitude of 400ft.

Thanks for your detailed analysis endotherm.  After RTH I did not used the sticks since I could not see anything but the ups and downs of the gimbal and I expected the P3P was going to return home but the it got into the spirals and crashed. I notice there is a displacement in the data you show in the photo, your path is much into the hiils than actually was, here you can see it in the phantomhelp (http://www.phantomhelp.com/LogViewer/9MWG6F524XQ655PSQUDK/#) look at the loop it is on the road and it yours is up into the hills.
I don't think there were any birds, I was at aprox 3500 mts over sea level and you don't see any in that place or around. Wind was gentle at no more that 3 m/s.
I use an Ipad Air2 and can't find a way to tell what version of DGI GO is installed, no way to get into the app since it can't communicate to the drone. Always keep the last firmware installed in the P3P, RC and batteries, also no way to tell you that.
I am in the process to getting the flight log getting analyzed by DJI, As I was told, even they don't do that out of warranty. They only do it if you send the aircraft first; I asked to please make a consideration since it would be a waste of money having to spend a couple of hundred $ sending and getting it back to Venezuela, plus the repair costs involved. I am still thinking it it worth spending money on a Phantom 3 Pro.
By the way, I am the same pilot with the P4P crash in the first flight at 2 mt heigh (https://forum.dji.com/forum.php? ... &fromuid=208349) and also thanks for your comments over there.
I am also including actual photos ot the P3P. Damages are: Broken Gimbal, broken camera cables, cracked hull, lost minor covers on legs and leds, and, the main circuit (I open the hull) has the 2 main connectors broken, not shown in photos)

https://www.dropbox.com/s/ye74xa ... 430_153824.jpg?dl=0
https://www.dropbox.com/s/rsrw6u ... 430_154014.jpg?dl=0

Might be worth a DIY repair, or buy the parts and get a local electronics expert to do it.  I think sending it in is unlikely to be cheaper than buying a new P3P (or P4 standard) if you can find one, they are out of production now so are cheap if you can still find one, your postage for a repair is not going to be cheap.

If you really want a 2nd Phantom then I suggest a P4A since it has the same great camera as your P4P, significantly better than your P3, so will be a decent backup aircraft and I believe they will be DJI's cheapest Phantom once they have sorted their latest updates out.  A DIY repair of the P3 will be significantly cheaper though.

Nigel_ Posted at 2017-5-1 04:46
Might be worth a DIY repair, or buy the parts and get a local electronics expert to do it.  I think sending it in is unlikely to be cheaper than buying a new P3P (or P4 standard) if you can find one, they are out of production now so are cheap if you can still find one, your postage for a repair is not going to be cheap.

If you really want a 2nd Phantom then I suggest a P4A since it has the same great camera as your P4P, significantly better than your P3, so will be a decent backup aircraft and I believe they will be DJI's cheapest Phantom once they have sorted their latest updates out.  A DIY repair of the P3 will be significantly cheaper though.

I agree Nigel. I am just waiting to see if DJI has any consideration and maybe it results in a cheap repair. I know this has happened before; that is why I insisted so much on then and they are going to review the log flight despite is out of warranty. I will post any news. On the other hand I still have to send the crashed P4P for warranty repair, you know this case from my other thread.

apomares-Vzla Posted at 2017-5-1 04:26
Thanks for your detailed analysis endotherm.  After RTH I did not used the sticks since I could not see anything but the ups and downs of the gimbal and I expected the P3P was going to return home but the it got into the spirals and crashed. I notice there is a displacement in the data you show in the photo, your path is much into the hiils than actually was, here you can see it in the phantomhelp (http://www.phantomhelp.com/LogViewer/9MWG6F524XQ655PSQUDK/#) look at the loop it is on the road and it yours is up into the hills.
I don't think there were any birds, I was at aprox 3500 mts over sea level and you don't see any in that place or around. Wind was gentle at no more that 3 m/s.
I use an Ipad Air2 and can't find a way to tell what version of DGI GO is installed, no way to get into the app since it can't communicate to the drone. Always keep the last firmware installed in the P3P, RC and batteries, also no way to tell you that.

You're welcome.

The data isn't actually displaced, it is parallax error. Normally when I upload these screen captures it is obvious it was taken from a 3D model, not from directly overhead looking down.  Not so in this case it seems.  This was captured from Google Earth in 3D and the view is from the side looking down at probably 45°.  I was trying to show the ups and downs in the course as well.  If you follow the thin white lines from the drop pins down through the flight path to where they intersect on the ground, that will show where the flight path was overhead.  The white lines are perfectly vertical.  Just imagine the yellow track floating over the road and the illusion should become clear.

In these grabs, the grey line is the shadow it would cast if the sun were directly overhead.  The second one is a low view from a different angle.


Again, I have no explanations for you as to what happened.  I am just speculating various scenarios that could explain this weird behaviour.  Winds along mountain faces is quite common though, even though the air remains still at ground level.

If you have updated your app this year, I don't think that bug would affect you, I believe it was confined to Android-only anyway.

I found your other thread via a link you posted earlier in this one, glad to help.  Trying to determine why these things happen actually helps everyone to fly safer.

The external damage doesn't look too bad;  it will still fly with twisted arms, and with new props and a new camera gimbal it would be usable.  It's a DIY job if you are capable.  However the camera assembly isn't that cheap, neither is the main board.  DJI would likely replace the shell, transplant all the internals and supply a new camera, as well as any other minor features.  It would be easier and cheaper for them to supply a refurbished replacement.  They might be generous and offer you a percentage discount as they sometimes do.  A replacement aircraft-only would also work out cheaper as you are not paying for a new remote and a new battery.  Just so you know what to expect.

endotherm Posted at 2017-5-1 10:37
You're welcome.

The data isn't actually displaced, it is parallax error. Normally when I upload these screen captures it is obvious it was taken from a 3D model, not from directly overhead looking down.  Not so in this case it seems.  This was captured from Google Earth in 3D and the view is from the side looking down at probably 45°.  I was trying to show the ups and downs in the course as well.  If you follow the thin white lines from the drop pins down through the flight path to where they intersect on the ground, that will show where the flight path was overhead.  The white lines are perfectly vertical.  Just imagine the yellow track floating over the road and the illusion should become clear.

Man, that is impressive!
I really like your detailed explanations. A lot to learn from.

Now here is your next challenge: Find Malaysia Airlines flight MH370!
That would make you immortal.

endotherm Posted at 2017-5-1 10:37
You're welcome.

The data isn't actually displaced, it is parallax error. Normally when I upload these screen captures it is obvious it was taken from a 3D model, not from directly overhead looking down.  Not so in this case it seems.  This was captured from Google Earth in 3D and the view is from the side looking down at probably 45°.  I was trying to show the ups and downs in the course as well.  If you follow the thin white lines from the drop pins down through the flight path to where they intersect on the ground, that will show where the flight path was overhead.  The white lines are perfectly vertical.  Just imagine the yellow track floating over the road and the illusion should become clear.

Thanks again endotherm. The second image shows the flight path perfectly. Sorry I did not notice it was in Google Earth, that happens to me all the time. Mountains are much smaller than it looks in the image but is is ok.

I don't care if DJI offerst a refurbished P3P at a good affordable price, that is what I am expecting since I should also pay for export to USA, import to Venezuela, customs and taxes as well.

You have done a great job and I am very grateful for all your help. I will let you know how my two crash cases with DJI end up.

apomares-Vzla Posted at 2017-5-1 11:06
Thanks again endotherm. The second image shows the flight path perfectly. Sorry I did not notice it was in Google Earth, that happens to me all the time. Mountains are much smaller than it looks in the image but is is ok.

I don't care if DJI offerst a refurbished P3P at a good affordable price, that is what I am expecting since I should also pay for export to USA, import to Venezuela, customs and taxes as well.

Mountains are much smaller than it looks in the image but is is ok.

Yes, that is just me.  I have the elevation exaggeration cranked up to 2, it makes for better pictures.    The terrain features are more noticeable that way.  It does keep 3D models and elevations relative and to scale, but I guess something is lost when adding a kml track.

Here is exaggeration = 0.5. Is this more natural?  The flight path doesn't look as vertically stretched either, so I guess everything is kept relative, just distorted in height.

Eric13 Posted at 2017-5-1 11:02
Man, that is impressive!
I really like your detailed explanations. A lot to learn from.

No problems.  Now if you could just supply me with the kml data and the flight log...