I've flown RC aircraft on and off over the years and I've flown my Mavic about 20 times so far.  I'm no stranger to keeping my "mind’s eye" on the orientation of the craft and responding with the correct control inputs.  But you don't have to believe me on this one, because this crash is so quick and obvious that there is no doubt something went wrong with the drone.

  
  

This is a video of the flight replay.

https://www.youtube.com/watch?v=IDNeOALmSZs

  
  

Here if the video of the flight to prove the drone was facing away from the pool.

  

https://www.youtube.com/watch?v=ibjq7JP8kiI

  
  

The flight replay shows that the drone was facing East (towards the pool).  In reality, it was facing West (away from the pool).  I'm pretty sure this is related to why it crashed.

  
  

At 10 seconds in you can see the drone drifting North.  This really surprised me because it’s usually dead steady.  I thought maybe it was just a minor issue with GPS.  I issue some small controls to bring the drone steady again, to avoid some plants.

  
  

At 20 seconds in you can see that the drone starts drifting South.  Unfortunately, this is exactly when the controller is alerting me that I'm in restricted airspace.  I have to click accept before continuing.  (There is a tiny airport a few miles away, I have permission from them to fly).

  
  

Before I even click the button, I notice the drift.  I immediately go full right stick.  I'm thinking, this is really weird, why can't it hover.  Within a second I know I'm going down.  The drone doesn't change attitude at all.  It's completely ignoring the right stick.

  
  

I also tried some right yaw, the drone did listen to that.  It wasn't much help though.

  
  

I've seen a few stories like mine on forums.  Most people just respond with "pilot error", "read the manual", "not safe to fly there", "too much interference".  These responses are inconsiderate to those having these problems.  Yes, pilot errors do happen.  But software/hardware errors happen too.  The drone is controlled by incredibly complicated pieces of software.  Software has bugs in it, it’s just the way it is.  Something is happening in the Mavic that causes it to drift and ignore control input.

  
  

I'm bummed it crashed.  The camera, gimbal, and props are broken.  I can't even get any camera images on the app, I probably need to replace the whole assembly.  I kept replaying the events in my head.  "Did I bump the stick left?"  "Did I get flipped around in my head?"  I'm glad I learned how to view the recordings (didn't even know that was a feature).  I was glad to see my stick inputs were correct.

  
  

Was not glad I broke my drone though.

P.S.  At least it didn't go into the pool.  It colider with the slide and fell just barely on concrete.

Did you have downward postioning sensors on? I usually turn mine off when hand catching, then when i take off from the ground i notice it drifts until i turn downward postioning sensors back on.......

Danjahmouse Posted at 2017-7-6 21:47
Did you have downward postioning sensors on? I usually turn mine off when hand catching, then when i take off from the ground i notice it drifts until i turn downward postioning sensors back on.......

I've never turned off the downward sensors.  The first drift and the beginning of the second one could be explained by something with the downward sensors.  But the last drift was getting pretty aggressive.  Even if it was a sensor issue, it didn't respond to my left/right stick.

This looks very much like a compass error. That is why it displays the wrong orientation. GPS seems to be just fine, but needless to say, the drone gets very confused. Ideally it should enter ATTI-mode whenever it detects that something is off, but unfortunately it is very hard for the systems to detect this kind of error. (This thread is a good read: https://mavicpilots.com/threads/compass-calibration-guide.12/ )

copy/paste form this thread:

"Warning Signs
The Mavic can only detect when the compass is providing extremely poor (implausible) data. This typically occurs if you place it near a strong magnetic field or do not calibrate it properly. It will flash red and yellow lights and the Mavic will indicate a compass error in the app.

IMPORTANT: The lack of a compass error does NOT mean your compass is working and calibrated properly"

Having seen this issue a lot in different forums, I always do check that the drones sensed orientation matches its real orientation before take off, and just after.

On this matter, the official DJI Mavic manual actually contradics what i understand to be best practic. The manual says to only calibrate compass whenever prompted.

fansd51eea3b Posted at 2017-7-6 23:24
This looks very much like a compass error. That is why it displays the wrong orientation. GPS seems to be just fine, but needless to say, the drone gets very confused. Ideally it should enter ATTI-mode whenever it detects that something is off, but unfortunately it is very hard for the systems to detect this kind of error. (This thread is a good read: https://mavicpilots.com/threads/compass-calibration-guide.12/ )

copy/paste form this thread:

... and sorry for your crash. :-( Looks like something that could have happened to anyone.
I am still very new to flying drones, and I am trawling the internet for crash-incidents to try to learn from them.

The off-compass does not really explain your difficulties with the stick inputs, but from what i have seen, one error is usually followed by several others.I guess the highly advanced software calculations are prone to behave incorrectly when all the different inputs mismatch.

miebster, we are sorry for the crash accident.  Please contact our support to raise a ticket and send in the drone, our data analysis team will help to figure out the reason of crash, and the further resolution will be provided after that, Here is the link to obtain support: http://www.dji.com/support

If you'd like to upload the DAT file from the flight I or someone else will look at it and may be able to shed some more light on why this happened. The instructions for accessing the DAT files are here and once you've retrieved the file you'll need to upload it to DropBox (or something similar) and post a link to it in this thread.

Most likely compass error.  Notice that the Arrow indicating the orientation of the aircraft is not the direction of flight.  You should have gotten a warning.  Can you post the  DAT file following instructions provided by DroneFlying?  Also the unedited video file.

DJI Mindy Posted at 2017-7-7 01:01
miebster, we are sorry for the crash accident.  Please contact our support to raise a ticket and send in the drone, our data analysis team will help to figure out the reason of crash, and the further resolution will be provided after that, Here is the link to obtain support: http://www.dji.com/support

I've created a ticket, thank you.

DroneFlying Posted at 2017-7-7 04:19
If you'd like to upload the DAT file from the flight I or someone else will look at it and may be able to shed some more light on why this happened. The instructions for accessing the DAT files are here and once you've retrieved the file you'll need to upload it to DropBox (or something similar) and post a link to it in this thread.

I wasn't able to get DJI Assistant to work.  The app in the google play store and apple store doesn't look like the one from the article.  Is there some other way to get the DAT file?

Griffith Posted at 2017-7-7 06:05
Most likely compass error.  Notice that the Arrow indicating the orientation of the aircraft is not the direction of flight.  You should have gotten a warning.  Can you post the  DAT file following instructions provided by DroneFlying?  Also the unedited video file.

What do you mean by unedited video file?  The camera video that I uploaded to you tube was the entire video, no editing.  The video showing the control inputs was from screen capturing the DJI Go app.

miebster Posted at 2017-7-7 07:52
What do you mean by unedited video file?  The camera video that I uploaded to you tube was the entire video, no editing.  The video showing the control inputs was from screen capturing the DJI Go app.

You can upload it here http://www.phantomhelp.com/LogViewer/Upload/ just click link and follow instructions, then come back here and post the link.
It will give you some understanding of what may have happened.

I think I found the right DAT file.

https://drive.google.com/open?id=0By3XDei-nRVKUjlfX2RrODliZGc

miebster Posted at 2017-7-7 07:52
What do you mean by unedited video file?  The camera video that I uploaded to you tube was the entire video, no editing.  The video showing the control inputs was from screen capturing the DJI Go app.

Sounded like you had dubbed the audio commentary.  I was curious to hear the RC commentary :-)

miebster Posted at 2017-7-7 07:50
I wasn't able to get DJI Assistant to work.  The app in the google play store and apple store doesn't look like the one from the article.  Is there some other way to get the DAT file?

Try downloading it from this page, which contains links to both Windows and Mac / iOS versions.

miebster Posted at 2017-7-7 08:16
I think I found the right DAT file.

https://drive.google.com/open?id=0By3XDei-nRVKUjlfX2RrODliZGc

No, that isn't the correct file. The correct one will have an extension of .DAT and will probably be in at least the tens of megabytes in size.

DroneFlying Posted at 2017-7-7 10:57
No, that isn't the correct file. The correct one will have an extension of .DAT and will probably be in at least the tens of megabytes in size.

This might be the right log file.

https://drive.google.com/file/d/ ... U0/view?usp=sharing

This website (http://www.phantomhelp.com/LogViewer/Upload/) gives an error when you click "upload log".

miebster Posted at 2017-7-8 18:19
This might be the right log file.

https://drive.google.com/file/d/0By3XDei-nRVKbks2NjVZeGhMWU0/view?usp=sharing

Yes, that's the correct DAT file. The reason you got an error uploading it is that PhantomHelp only accepts TXT log files (from the mobile device) but I was able to review it using BudWalker's CsvView.

Unfortunately this looks very much like a case where takeoff occurred from a geomagnetically distorted location. In other words, something near the aircraft -- I'd guess rebar within the concrete around your pool -- distorted the compass heading. That bad compass heading (magYaw) was used to set what the aircraft believed to be its current heading (yaw), and as Paddyralph and Griffith suggested, that's why the replay in the Go app shows a different (incorrect) heading from what you knew to be the actual heading as shown in the video.

I've included a graph from CsvView below that should help somewhat to understand what occurred:

• The red plot is your height / altitude
• The dark green / turquoise plot is your aileron (move left / right stick input)
• The purple plot is your rudder (rotate left / right stick input)
• The light green line is yaw (expected heading)
• The blue line is magYaw (compass reading)
  

Those last two are the most important and as you can see right before takeoff yaw and magYaw are essentially identical, but once you lifted off magYaw diverged significantly. That's the compass getting out of range of the interference on the ground and the slight downward trend in magYaw you see is the flight controller (FC) rotating the aircraft left to try to get yaw and magYaw realigned, and about 9 seconds into the flight you applied aileron and then rudder input. The acceleration you saw right before the crash wasn't a result of any stick input by you, but was apparently caused by the FC trying to maintain position using the bad heading information and is pretty standard behavior for the Mavic in these circumstances. Frankly I suspect that you would have fared better with a newer version of the firmware than .0200 because at some point DJI updated the behavior to have the Mavic revert to ATTI mode, probably for specifically this kind of situation.

So the bottom line is that unfortunately this appears to have been caused by the choice of takeoff location. In this situation people will often comment that they've taken off from the same location before, because even a small change in position can have a big impact on the compass deviation and on what happens once you're airborne.

DroneFlying Posted at 2017-7-9 11:02
Yes, that's the correct DAT file. The reason you got an error uploading it is that PhantomHelp only accepts TXT log files (from the mobile device) but I was able to review it using BudWalker's CsvView.

Unfortunately this looks very much like a case where takeoff occurred from a geomagnetically distorted location. In other words, something near the aircraft -- I'd guess rebar within the concrete around your pool -- distorted the compass heading. That bad compass heading (magYaw) was used to set what the aircraft believed to be its current heading (yaw), and as Paddyralph and Griffith suggested, that's why the replay in the Go app shows a different (incorrect) heading from what you knew to be the actual heading as shown in the video.

I'm totally new to drones and will buy my Mavic tomorrow. Regarding this case I have one question to DroneFlying or others with knowledge.

If the original topic poster would have done a self initiated compass calibration before flight, would that be a difference in the outcome?

mozso Posted at 2017-7-9 12:26
I'm totally new to drones and will buy my Mavic tomorrow. Regarding this case I have one question to DroneFlying or others with knowledge.

If the original topic poster would have done a self initiated compass calibration before flight, would that be a difference in the outcome?

If the original topic poster would have done a self initiated compass calibration before flight, would that be a difference in the outcome?

Probably not. The key lesson from this case is to make sure that you aren't taking off from a place where there's some sort of geomagnetic distortion. Concrete intuitively seems like a good surface from which to launch but many people don't realize that it usually contains (steel) rebar which can affect the Mavic's compass and result in erratic flight like that shown here.

Makes sense, is there any sign coming from drone instruments indicating such compass anomaly before takeoff while the drone is powered on and ready to fly? Or should I buy a magnetic anomaly detector also to scan the take off spot?

mozso Posted at 2017-7-9 12:26
I'm totally new to drones and will buy my Mavic tomorrow. Regarding this case I have one question to DroneFlying or others with knowledge.

If the original topic poster would have done a self initiated compass calibration before flight, would that be a difference in the outcome?

Probably not.
When you calibrate the compass you take the drone in your hand. Just that can change the calibration.
the floar when the mavic is can have electromagnetic and just puting in the air, in your hand can change that.

Always verify the sensor status and also take a look to the map to see the orientation of the mavic.

DroneFlying Posted at 2017-7-9 11:02
Yes, that's the correct DAT file. The reason you got an error uploading it is that PhantomHelp only accepts TXT log files (from the mobile device) but I was able to review it using BudWalker's CsvView.

Unfortunately this looks very much like a case where takeoff occurred from a geomagnetically distorted location. In other words, something near the aircraft -- I'd guess rebar within the concrete around your pool -- distorted the compass heading. That bad compass heading (magYaw) was used to set what the aircraft believed to be its current heading (yaw), and as Paddyralph and Griffith suggested, that's why the replay in the Go app shows a different (incorrect) heading from what you knew to be the actual heading as shown in the video.

Thank you for this little investigation DroneFlying, very helpful, as i've experienced something like this, and i to was taking of from a concrete platform, and could'nt undestand why it happened, mine drone - thank god - did'nt crash, so i was lucky.., i never started from that location again, and after your post here, nore from any concrete-platform

I have one question. If the drone is placed in a geomagnetically distorted location, what is the reason of that the magYaw and Yaw diverse right after the take-off rather than before take off?

isaacchu Posted at 2017-7-10 01:04
I have one question. If the drone is placed in a geomagnetically distorted location, what is the reason of that the magYaw and Yaw diverse right after the take-off rather than before take off?

The rapidly changing sensor readings.

I see that it’s mentioned that the new firmware should make this safer. Do anyone know how reliable the new firmware is to alert us of such interference and problem that’s been mentioned? I think it’s essential that we know the safety regard this. The other day I got no warnings about compass error before flight, but about 50m away from me the drone start traveling to the side with the tale light flashing red in a rapid manner, than going over to ATTI mode for 1-2 seconds and back to GPS mode. After landing it I found in the compass menu that I should recalibrate.. why didn’t it warn me straight away? before take-off?  FYI. Used current firmware. Drone and DJI GO app.

Does most people recalibrate the compass every time before  a flight?

mozso Posted at 2017-7-9 22:13
Makes sense, is there any sign coming from drone instruments indicating such compass anomaly before takeoff while the drone is powered on and ready to fly? Or should I buy a magnetic anomaly detector also to scan the take off spot?

is there any sign coming from drone instruments indicating such compass anomaly before takeoff while the drone is powered on and ready to fly?

Yes, you should receive a message that reads, "Move aircraft or calibrate compass" when the aircraft is located in a geomagnetically distorted location before takeoff. I've seen the message many times and have never had a problem as long as I moved the Mavic when told to do so.

isaacchu Posted at 2017-7-10 01:04
I have one question. If the drone is placed in a geomagnetically distorted location, what is the reason of that the magYaw and Yaw diverse right after the take-off rather than before take off?

what is the reason of that the magYaw and Yaw diverse right after the take-off rather than before take off?

The yaw value is what the flight control (FC) thinks the current heading is / should be based on things like rotation of the aircraft and is initially set to the magYaw (compass) value. Once takeoff occurs and the aircraft is free of magnetic interference the compass automatically corrects itself but the yaw value -- which was a copy of the bad compass value -- remains unchanged.

Fixaldo Posted at 2017-7-10 05:36
I see that it’s mentioned that the new firmware should make this safer. Do anyone know how reliable the new firmware is to alert us of such interference and problem that’s been mentioned? I think it’s essential that we know the safety regard this. The other day I got no warnings about compass error before flight, but about 50m away from me the drone start traveling to the side with the tale light flashing red in a rapid manner, than going over to ATTI mode for 1-2 seconds and back to GPS mode. After landing it I found in the compass menu that I should recalibrate.. why didn’t it warn me straight away? before take-off?  FYI. Used current firmware. Drone and DJI GO app.

Does most people recalibrate the compass every time before  a flight?

To be clear, the "newer" firmware I was mentioned (assuming that's what you were referring to) was anything later than the .0200, which is a very old version in terms of the Mavic's short lifetime. And yes, I believe the newer firmware versions handle this situation reasonably well.

The experience you described might have been caused by taking off from a geomagnetically distorted launch site, but based on your description I'd guess that wasn't the problem. There are other scenarios where ATTI mode can be triggered and I'd guess that was one of them.

I don't believe most people recalibrate their compass before every flight, but in any case DJI doesn't recommend that and neither do I.

DroneFlying Posted at 2017-7-10 06:47
is there any sign coming from drone instruments indicating such compass anomaly before takeoff while the drone is powered on and ready to fly?

Yes, you should receive a message that reads, "Move aircraft or calibrate compass" when the aircraft is located in a geomagnetically distorted location before takeoff. I've seen the message many times and have never had a problem as long as I moved the Mavic when told to do so.

Thanks for your explanation. I did have a similar issue when I try to take off on a concrete floor (probably having rebars underneath). The DJI app remind me there was a strong magnetic interference and not allow me to take off. Since then, I do not take off on concrete surface anymore. Do you have any idea in what circumstance the DJI app can not detect this kind of magnetic interference and allow me to take off? I certainly do not want to crash my drone like this. Thanks very much.

isaacchu Posted at 2017-7-10 07:36
Thanks for your explanation. I did have a similar issue when I try to take off on a concrete floor (probably having rebars underneath). The DJI app remind me there was a strong magnetic interference and not allow me to take off. Since then, I do not take off on concrete surface anymore. Do you have any idea in what circumstance the DJI app can not detect this kind of magnetic interference and allow me to take off? I certainly do not want to crash my drone like this. Thanks very much.

Do you have any idea in what circumstance the DJI app can not detect this kind of magnetic interference and allow me to take off?

No, I couldn't really identify any specific circumstances where it wouldn't be detected, though that doesn't mean there aren't any. I think the important thing is just to use good judgment in choosing your takeoff location and to move the aircraft when requested.

I had an instance this weekend while shooting indoors where I landed and the Mavic did a quick yaw on touch down. We were in a large convention center on the second floor, so I am sure there is a lot of steel reenforcement. Other than that, I flew roughly nine batteries this weekend, indoors, without GPS, and she was rock solid. Often I would take off and land on a table to avoid the magnetic interference.

DroneFlying Posted at 2017-7-9 11:02
Yes, that's the correct DAT file. The reason you got an error uploading it is that PhantomHelp only accepts TXT log files (from the mobile device) but I was able to review it using BudWalker's CsvView.

Unfortunately this looks very much like a case where takeoff occurred from a geomagnetically distorted location. In other words, something near the aircraft -- I'd guess rebar within the concrete around your pool -- distorted the compass heading. That bad compass heading (magYaw) was used to set what the aircraft believed to be its current heading (yaw), and as Paddyralph and Griffith suggested, that's why the replay in the Go app shows a different (incorrect) heading from what you knew to be the actual heading as shown in the video.

Thanks for this analysis.  Its actually quite alarming to realize how much the Mavic depends on compass information.  Its also alarming that the vision sensors didn't keep it hovering in place.

DJI Mindy Posted at 2017-7-7 01:01
miebster, we are sorry for the crash accident.  Please contact our support to raise a ticket and send in the drone, our data analysis team will help to figure out the reason of crash, and the further resolution will be provided after that, Here is the link to obtain support: http://www.dji.com/support

DJI Support responded with

"Dear Customer, Unfortunately, damage that is not caused by a product malfunction or is out of the warranty period is not covered by DJI aftersales policy. We will either repair it or replace it with a product that's new or equivalent to new in both performance and reliability after payment has been received. For more information, please visit http://www.dji.com/service/policy DJI North America"

The drone ignoring my stick inputs isn't a a product malfunction?

miebster Posted at 2017-7-19 11:54
DJI Support responded with

"Dear Customer, Unfortunately, damage that is not caused by a product malfunction or is out of the warranty period is not covered by DJI aftersales policy. We will either repair it or replace it with a product that's new or equivalent to new in both performance and reliability after payment has been received. For more information, please visit http://www.dji.com/service/policy DJI North America"

Please leave me with your case number so that I can check more details, thanks.