Yesterday, I wanted to give the spark a fly yesterday and was first forced to update the firmware on the aircraft and the controller.

Once the software was all updated, it was time for take off! Home point set, no compass or any other issues

After climbing directly from takeoff, sparky decided it was time for an adventure and flew off at nearly 40km/h!

At this point the aircraft was no longer reacting to inputs i was making on the controller. RTH timed out. It was basically out there on its own!

According to the go app, it came down (and disconnected immediately) on the roof of a large casino. However, i saw continue to fly well beyond this building until it was out of site.  But, to satisfy my curiosity, i went in to the casino, and someone there kindly took me out on to the roof where we searched for nearly an hour!

Here's a video of the final 1:28 seconds of my sparks life! After take off it reported a compass error, and that was the point it left my control.



and the search....

Save

Save

no idea why the pictures are in this orientation now ::

I'm sorry to hear that. Please contact our support and start a case: https://www.dji.com/support?site=brandsite&from=nav You can provide your flight records to our support for further analysis. Thanks for your support.

Thanks.
I sent a support request yesterday already. Still waiting for a reply.

Dang - I'm sorry to hear that.  It looks like there were a lot of wind warnings.  Maybe your Spark was fighting the wind too hard which sent it into ATTI mode (just a thought)?  And then drifted to it's final end point.  

Antpal Posted at 2018-7-24 06:50
Thanks.
I sent a support request yesterday already. Still waiting for a reply.

Antpal,

Please upload your flight record to http://www.phantomhelp.com/LogViewer/Upload/ and share the link with us.
We could analyze what happened.
From your short video it looks like you had problems with Spark compass calibration from the beginning and that Spark went into ATTI because of that and was blown away by the wind.

Your flight record may help us to advise you were to look for it and to figure out what happened.

Mirek

Bender1031 Posted at 2018-7-24 07:04
Dang - I'm sorry to hear that.  It looks like there were a lot of wind warnings.  Maybe your Spark was fighting the wind too hard which sent it into ATTI mode (just a thought)?  And then drifted to it's final end point.

Hi,

There was hardly any wind at all.
There was also no drifting involved from what i saw (i had it in eye sight the whole flight). It flew at full speed and coninuted to climb. And i was helpless to do anything.

Antpal Posted at 2018-7-24 07:16
Hi,

There was hardly any wind at all.

From the Video you have Warnings of High Wind Velocity, thats why many are telling you there was strong winds.

Remember that the Higher you go, the Stronger the Force will be.

As others told you it went in ATTI and drifted away. It looks like it was flying alone, thats why most pilots call it 'Fly Away'.

Upload the FlightLog when you can so others can help you out.

DMX_MT Posted at 2018-7-24 07:20
From the Video you have Warnings of High Wind Velocity, thats why many are telling you there was strong winds.

Remember that the Higher you go, the Stronger the Force will be.

Looking in to uploading it now. Thanks.

As strange as it sounds, there really was very little wind. I've flown in windy conditions without problems.

Even on the roof of the building it apparently crashed on, there was very little breeze.


I'm now a couple of hundred KM away from the location, so the chances of me finding it are now nil. If someone comes across it, it has a registration plate on it.

The horizontal speed is also confusing. Once the location indicator reaches the roof of the casino, it's still showing horizontal speed.

But, from the point it started climbing rapidly, no inputs from the controller (even starting video) didn't work. The aircraft simply no longer reacted on my inputs.

Hi,

Here is the flight data - what a shock!!! It did fly much beyond the casino roof as i could see with my own eyes! Why did the DJI app not deliver this information?? I may have stood a chance of finding the thing!!!

http://www.phantomhelp.com/LogViewer/KH1YHA9KFSVOG0TB1PI2/

What the hell is a Speed Error ?

In Flight, working IMU encounters heading exception,please switch to atti mode if craft behave abnormally; Yaw Error


Looks like this is a DJI problem. I would put up a refund request dude!

Simple way to avoid being "forced" to update the app/firmware is to put your device in airplane mode before doing any droning.  I have an iPad Mini dedicated to the Spark and it's always in airplane mode so I only update when I'm ready to.

Yeah, the amount of errors on screen were crazy. And I got to the point that i was operating the controller like it was from a playstation!! No response.

And since ive seen the flight txt file, I'm totally p***ed that I was looking in completely the wrong location for the thing.

Antpal Posted at 2018-7-25 00:17
Hi,

Here is the flight data - what a shock!!! It did fly much beyond the casino roof as i could see with my own eyes! Why did the DJI app not deliver this information?? I may have stood a chance of finding the thing!!!

Antpal,

  

Your Spark displayed serious compass errors from the beginning.

Also, your Spark did react correctly to your stick inputs – there was no problem there. You were only raising it up and it did respond correctly to your instructions to gain altitude.

  

Here is a bit of analysis:

  

• 7.1s: First compass error. When you see something like this you should land immediately (if it is safe to do so).
• 15.4s: You are raising altitude but Spark cannot hold the same GPS position due to serious compass issue – first Speed Error (this indicates GPS position mismatch).
  

  

You should have noticed that Spark is slowly flying away without you instructing it to do so. Instead you keep raising altitude.

  

• 27.9s: First high velocity wind warning. Spark is already far away from you without you instructing it to do so. Perhaps you did not notice that.
• 32.4s: You stop raising altitude. Spark is almost 150 metres horizontally from you and already experiencing toilet bowl effect which is caused by total confusion and mismatch between compass, accelerometer and GPS data. The altitude is over 52 metres.
• 36.5s: You finally react to abnormal behaviour and press RTH. This will not help. Spark is in seriously toilet bowling and seriously confused. RTH has no chances of success. Dropping to ATTI is imminent.
• 42.1s: Spark’s firmware gives up and drops to ATTI to allow you to regain manual control. This is correct behaviour. You do nothing since you probably lost sight of Spark already.
  

      

Spark accelerates being pushed by the wind in SE direction. Wind was very gusty (varying speed)

  

Even if you looked for Spark at the end of the red line, you would not have found it there. When communication broke, Spark was at the altitude of 53 metres flying away with the speed of 4m/s.

  

3 seconds after communication broke, Spark would enter Auto-Landing sequence and start to lower its altitude with a speed of 2.6 to 3 m/s. The buildings on its path are probably no higher than 10metres. Between end of red line and the point were Spark was at 10 metres altitude required about 16-20 seconds. Spark would move between 60 and 80 metres during this time (pushed by the wind).

  

See picture below for a possible crash zone location.

  

So what happened?

  

There are two likely scenarios:

  

• You started Spark from a location close to heavy magnetic interference. Yes, I know it was on a beach, but you could have been in a close proximity to some metal structure. Spark warned about compass trouble few seconds into the flight. You ignored the warning. If Spark was starting with magnetic interference, while it was raising up the interference would weaken causing compass to “unwind” and causing a serious mismatch between Spark’s physical heading and compass direction.
• Compass was not calibrated properly.
  

  

There is third one – h/w o firmware issue – but this is much less likely.

  

Since you lost your Spark, I can only provide advice to anybody reading this – as a future warning. If you see compass exception or compass error, stop your flight and land as soon as safely possible. In this case you ignored warnings and you kept raising Spark up while watching it to fly away. You did that for 40 seconds consistently without any effort to either land or bring it back.

  
I am sorry for your loss.  

Mirek

Thanks for the info mirek6.

Given the compass error, it had already to started to fly without any import from me. By the time I'd realised it wasn't climbing at my command, but on its own, the probably would have been little I could do anyway.

The buildings in the area of the crash zone were all 3 stories or more - all over 10 meters, most closer to thirty.

Wind, regardless of anything, would not have been an issue. There simply wasn't any else I wouldn't have flown.
Even on the roof of the casino, 40 meters up, i could have let go of a feather and it would have dropped to my feet.

DJI have checked the data and confirmed a hardware error, which will be dealt with under warranty  

Antpal Posted at 2018-7-25 07:45
Thanks for the info mirek6.

Given the compass error, it had already to started to fly without any import from me. By the time I'd realised it wasn't climbing at my command, but on its own, the probably would have been little I could do anyway.

Antpal,

  

  

Thanks for your reply. It is good to know that DJI concluded h/w issue – I am not surprised although I did consider it less likely scenario (more likely being calibration issues).

  

  

Yes – it is evident from the logs that you did not provide any directional input to Spark (your aileron and elevator inputs where neutral all the time), yet the Spark was flying away from you to the South/West. You did, however, provide throttle input (raising Spark to 53 metres) – at least this is what logs show and they show that Spark stopped raising as soon as you eased on throttle.

  

  

Flying on its own to SW was not caused by wind. This was spontaneous and erroneous behaviour of confused Spark.

  

The big circle which follows is a beautiful (and rare – for Spark) example of toilet bowl effect when AC continuously adjusts its bearings due to mismatch between commanded and physical positions of the drone. Spark knew that its commanded position remains home point and, after flying away, it was trying to go back but compass bearings were inconsistent with Spark direction.

  

  

Your decision to press RTH was a death sentence to Spark. While flying home the mismatch became so pronounced that firmware dropped Spark to ATTI.

  

  

I do not question what you felt as far as wind is concerned, but at 50 metres it may have been very, very different than at ground level or even on top of casino. And, interpreting your logs, it was uneven and gusty. When you lost contact with Spark it looks as if it was being pushed by gusts of about 15 km / hour. This is not strong wind. It was from the direction of the sea, which is what I would expect. Also, in ATTI, Spark does turn off its directional input. Spark’s power consumption during ATTI fly-away dropped which suggests that there was no more power given to propellers that necessary to keep it afloat. Hence my interpretation of wind as a cause of ATTI fly-away.

  

  

Having said that. There is another possible explanation which may be corroborated by DJI decision to replace your Spark:

• No wind but Spark drifts to the SW due to h/w failure. Sensory input is interpreted erroneously by firmware as being pushed by very strong wind because Spark cannot keep its position even though there is no horizontal input from sticks. With very strong winds (perhaps around 50 km/h) Spark will fail to keep its position in GPS mode. This is why, even though there was no wind, you got high velocity wind warnings. Wrong interpretation of what is happening in reality.
• Drift in ATTI due to IMU fault. If IMU is un-calibrated or faulty, Spark may not be level. If it is not level and in ATTI, it will drift with no wind.
  

  

  

I guess we will never know but your input provided valuable new information to similar cases.

  

  

Thanks and good luck with new Spark.

  

When you start flying new Spark, always do a 30 sec pre-check and warmup before you send it off far away.

  

  

Mirek

Thanks for the very detailed reply.
Much appreciated

Mirek6 Posted at 2018-7-25 07:31
Antpal,
  Your Spark displayed serious compass errors from the beginning. Also, your Spark did react correctly to your stick inputs – there was no problem there. You were only raising it up and it did respond correctly to your instructions to gain altitude.  
Here is a bit of analysis:  

Ill go with this. good call, don't fly in wind

Glad your getting a replacement. Keep in mind, you dont have to do a firmware update. Just ignore if you want. Many do.

BIG thanks to DJI!!

6 day after my flyaway I have a new bird, complete with memory card.

Awesome customer service experience!

Lucky you!
And thanks, DJI!

This is interesting not least of which one should not assume it is most likely pilot error by going down the rabbit hole of data analysis. A good end to the story! Customer is sent a replacement, DJI have a set of data that should help in ameliorating the problem in the future.

Antpal Posted at 2018-7-24 07:16
Hi,

There was hardly any wind at all.

This looks like the exact same thing as mine, I could see it all the time, I lost control and it started to climb and then flew off at light speed until it was no longer visible. Even the error messages are similar.

Mirek6 Posted at 2018-7-25 10:32
Antpal,     Thanks for your reply. It is good to know that DJI concluded h/w issue – I am not surprised although I did consider it less likely scenario (more likely being calibration issues).     Yes – it is evident from the logs that you did not provide any directional input to Spark (your aileron and elevator inputs where neutral all the time), yet the Spark was flying away from you to the South/West. You did, however, provide throttle input (raising Spark to 53 metres) – at least this is what logs show and they show that Spark stopped raising as soon as you eased on throttle.     Flying on its own to SW was not caused by wind. This was spontaneous and erroneous behaviour of confused Spark.  The big circle which follows is a beautiful (and rare – for Spark) example of toilet bowl effect when AC continuously adjusts its bearings due to mismatch between commanded and physical positions of the drone. Spark knew that its commanded position remains home point and, after flying away, it was trying to go back but compass bearings were inconsistent with Spark direction.      Your decision to press RTH was a death sentence to Spark. While flying home the mismatch became so pronounced that firmware dropped Spark to ATTI.     I do not question what you felt as far as wind is concerned, but at 50 metres it may have been very, very different than at ground level or even on top of casino. And, interpreting your logs, it was uneven and gusty. When you lost contact with Spark it looks as if it was being pushed by gusts of about 15 km / hour. This is not strong wind. It was from the direction of the sea, which is what I would expect. Also, in ATTI, Spark does turn off its directional input. Spark’s power consumption during ATTI fly-away dropped which suggests that there was no more power given to propellers that necessary to keep it afloat. Hence my interpretation of wind as a cause of ATTI fly-away.     Having said that. There is another possible explanation which may be corroborated by DJI decision to replace your Spark:

No wind but Spark drifts to the SW due to h/w failure. Sensory input is interpreted erroneously by firmware as being pushed by very strong wind because Spark cannot keep its position even though there is no horizontal input from sticks. With very strong winds (perhaps around 50 km/h) Spark will fail to keep its position in GPS mode. This is why, even though there was no wind, you got high velocity wind warnings. Wrong interpretation of what is happening in reality.

Drift in ATTI due to IMU fault. If IMU is un-calibrated or faulty, Spark may not be level. If it is not level and in ATTI, it will drift with no wind.

Hi Mirek, just checking who’s side you are on.   These are not cheap drones and the s/ware should be robust enough to ensure fly always don’t happen in any condiotns or pliot experience.  My own spark did the same and made no attempt to auto land, it just climbed and then shot off at Vmax.

I appreciate Mirek6 input.
The two obvious things that I did in this case was climb - i was taking off, and any interference during this time would have in my opinion stopped once I climbed enough to clear surrounding buildings.
And secondly, of course when I realised things were going wrong i pressed RTH. Im not sure who wouldn't in that situation