I had a real scare today! I had taken the Spark to 126 metres to get pictures of my house roof to check something. Then I wander a around for a minute or two and RTH. The drone returned from about 15 metres away and then went into Landing mode. As it descended, suddenly the engines seemed to cut and it started to spin horizontally. When I checked the playback, it showed the descent reaching 4.5 metres/second. I tried to hit the STOP button to see if maybe its RTH was faulty but nothing happened and I was horrified to see it coming down fast.

Very, very fortunately again (second time this week but the other time was a battery low situation and it landed safely in my garden in time), it was right over my awning and it hit the awning went back up into the air and seemed to hover so I tried to regain control and it responded. I brought it back and landed safely. No damage visible but I don't know if anything inside got shaken loose. I have lost confidence now as I don't know what could happen next. Last weekend, I had signal issues which I read were due to some update but then the next time I flew, things seemed to be okay and that's why I flew again today. My software and firmware is up to date.

I would be interested to know what happened and if I can be told how to download the black box data to send to DJI, I would do so.

Assistance and advice would be appreciated.

Chips

Sorry to hear about your issue, have you tried uploading the flight data using the DJI Assistant 2 Application? It probably makes sense not to fly the Spark for now until you find out more as to what happened on that flight? I would probably try to send DJI the data and speak with them about the condition of your drone, etc.

davidmartingraf Posted at 2018-4-30 21:10
Sorry to hear about your issue, have you tried uploading the flight data using the DJI Assistant 2 Application? It probably makes sense not to fly the Spark for now until you find out more as to what happened on that flight? I would probably try to send DJI the data and speak with them about the condition of your drone, etc.

Thanks for your advice, David. I have downloaded the Black Box data using DJI Assistant 2 and the contents total 42MB. But I don't know who to send it to for studying what went wrong since I myself won't know what I am looking at even if I can see the codes.

You're right about not flying it now that it had an impact, even though it seems to still fly ok. I was very very lucky today and I have always been terrified of such a failure when I am over a public road, or it comes down in my neighbourhood and hits someone.

With such devices, I tend to lose all confidence and I will probably have to buy a new one or move on to a Mavic Air. It's a shame since I have bought a number of accessories in the past 6 months ( have had the Spark since last September) and if I go to a Mavic Air, it means spending money all over again. This is my third drone in 12 months - I started with a Phantom 3A and then lost it in trees, got a Spark and then felt it was somewhat unstable in strong winds, and went and got a Phantom 3SE which I don't fly so often because it's bulky to take around. DJI has certainly gotten a lot of money from me!

fans0c5f7811 Posted at 2018-4-30 22:07
Thanks for your advice, David. I have downloaded the Black Box data using DJI Assistant 2 and the contents total 42MB. But I don't know who to send it to for studying what went wrong since I myself won't know what I am looking at even if I can see the codes.

You're right about not flying it now that it had an impact, even though it seems to still fly ok. I was very very lucky today and I have always been terrified of such a failure when I am over a public road, or it comes down in my neighbourhood and hits someone.

I'm glad to know that I could be of some help. In terms of the Black Box data, you can upload it to DJI so you can open up a case with them to see what exactly went wrong from their expert opinion?

I can understand your dilemma having purchased all those drones within the past twelve months. I wouldn't get too discouraged, you shouldn't be forced in writing off the Spark as a less than 1 year investment.

Did you purchase DJI Care Refresh? If not, judging from the sound of it the repair process shouldn't cost too much as compared to your overall cost already.

I would hold off on the Mavic Air front, I know the weather is getting nice and you'd like to enjoy flying your drones, but you're right about the higher cost as the Mavic Air isn't cheap.

Best of luck.

The data from the AC or android/IOS device can be saved & converted via http://datfile.net/. Search. I don't know what the black box does. Maybe nothing. All the info is in the flight data from the device/RC/AC. I can't convert the black box data - seems redundant?

fans0c5f7811 Posted at 2018-4-30 22:07
Thanks for your advice, David. I have downloaded the Black Box data using DJI Assistant 2 and the contents total 42MB. But I don't know who to send it to for studying what went wrong since I myself won't know what I am looking at even if I can see the codes.

You're right about not flying it now that it had an impact, even though it seems to still fly ok. I was very very lucky today and I have always been terrified of such a failure when I am over a public road, or it comes down in my neighbourhood and hits someone.

Any interference around your spark? I would recommend you enter wifi setting and see if most channels are green. If you ensure that it is not caused by environment, you can export the flight data and the black box for analysis. You can upload it to dropbox and post the link here. Thanks for your support.

djiuser_RojfSW97etD9 Posted at 2018-4-30 22:48
The data from the AC or android/IOS device can be saved & converted via http://datfile.net/. Search. I don't know what the black box does. Maybe nothing. All the info is in the flight data from the device/RC/AC. I can't convert the black box data - seems redundant?

I downloaded the data off the Spark itself right away in case it gets overwritten if I do fly it again - which I probably won't as I would always be nervous wondering if it will happen again with this particular unit. I am very conscious of public safety - wiping out a drone in a crash or losing it is one thing but if it comes down on a highway or someone's head, that is really bad.  

I hope someone at DJI will take an interest and ask me for the data so they can examine what went wrong. I don't expect electronic systems to be 100% reliable and when there is a fault or failure, I just want to know why to understand what happened.

DJI Elektra Posted at 2018-4-30 23:14
Any interference around your spark? I would recommend you enter wifi setting and see if most channels are green. If you ensure that it is not caused by environment, you can export the flight data and the black box for analysis. You can upload it to dropbox and post the link here. Thanks for your support.
https://www.youtube.com/watch?v=ngEHdAlAh9A

Thanks for your interest in my experience. I will send the data later today as I am on the road now.

DJI Elektra Posted at 2018-4-30 23:14
Any interference around your spark? I would recommend you enter wifi setting and see if most channels are green. If you ensure that it is not caused by environment, you can export the flight data and the black box for analysis. You can upload it to dropbox and post the link here. Thanks for your support.
https://www.youtube.com/watch?v=ngEHdAlAh9A

Thanks for your interest in my experience. I will send the data later today as I am on the road now.

fans0c5f7811 Posted at 2018-4-30 23:24
Thanks for your interest in my experience. I will send the data later today as I am on the road now.

Thanks for your cooperation. We are willing to help.

You should thoroughly check all the props. Remove them and make sure they're not cracked at the hub or something. Rotate the motors with the props off and make sure none of them are binding or have resistance when you turn them.

Could also be your battery is faulty and not providing enough current to keep your Spark in the air. Does it happen with only one specific battery?

Firehawk989 Posted at 2018-5-1 11:04
You should thoroughly check all the props. Remove them and make sure they're not cracked at the hub or something. Rotate the motors with the props off and make sure none of them are binding or have resistance when you turn them.

Could also be your battery is faulty and not providing enough current to keep your Spark in the air. Does it happen with only one specific battery?

I removed the props and they don't look any different from the extra set I have.

The hubs don't have any visible cracks and rotate freely when I push against them.

According to the flight log, the battery was 96% charged at take-off and 6.5 minutes later when it was in LANDING mode, it was at 22% and remained at that level all the way to shutdown. Of course, what the log shows and what the battery was actually delivering is something I would not be able to know without equipment to run tests. The battery is not the first one I had and is the third one I purchased which was 4 months ago. My batteries are charged using the flat hub provided by DJI with the Spark kit. When not in use, they are stored in a battery pouch.

Sounds like a bad battery, should be longer than 6.5min to get to 22% from 96. I've seen a few reports of batteries bought separately being faulty even directly from DJI.

If you pull your flight log from the phone it should tell you what the cell voltages were doing throughout the flight. If you know which battery it was I would stop using that one and test the others just hovering at min height over some grass or a cushion of some kind to see what flight times you're getting. You could test the potentially bad one this way too.

Firehawk989 Posted at 2018-5-1 21:57
Sounds like a bad battery, should be longer than 6.5min to get to 22% from 96. I've seen a few reports of batteries bought separately being faulty even directly from DJI.

If you pull your flight log from the phone it should tell you what the cell voltages were doing throughout the flight. If you know which battery it was I would stop using that one and test the others just hovering at min height over some grass or a cushion of some kind to see what flight times you're getting. You could test the potentially bad one this way too.

Thanks! That's a good tip for testing the batteries. I will certainly do that this weekend.
Here's the flight log. If you could look at it, perhaps you can spot something unusual. There are two messages in red during the stage of falling but they don't seem to explain why the motors would shut down.
http://www.phantomhelp.com/LogViewer/KJPJHCX5FUI9G1HW1R6N/

fans0c5f7811 Posted at 2018-5-2 02:00
Thanks! That's a good tip for testing the batteries. I will certainly do that this weekend.
Here's the flight log. If you could look at it, perhaps you can spot something unusual. There are two messages in red during the stage of falling but they don't seem to explain why the motors would shut down.
http://www.phantomhelp.com/LogViewer/KJPJHCX5FUI9G1HW1R6N/

Your flight log looks good.

  i had the same GPS position mismatch together with the speed error.  But the RTH worked fine. This has happened once in three months or so.

So please continue flying .
One suggestion is hoover the SPARK until it drains to check the battery and then recharge to full capacity and do again until it shuts off by itself.

fans0c5f7811 Posted at 2018-5-2 02:00
Thanks! That's a good tip for testing the batteries. I will certainly do that this weekend.
Here's the flight log. If you could look at it, perhaps you can spot something unusual. There are two messages in red during the stage of falling but they don't seem to explain why the motors would shut down.
http://www.phantomhelp.com/LogViewer/KJPJHCX5FUI9G1HW1R6N/

I've had a look at your log file and I believe the messages in red (Speed Error and GPS Position NoMatch) are very likely AFTER the fall when it hit your awning.

Looking at the log file, starting somewhere close to around the 6m 50s mark, we can see behavior that does not look normal at all.  Around that time frame (and it gets worse in the following seconds) we can see that the descent speed goes from just under 3m/s (which is the documented max descent speed for the Spark) to over 6m/s at around the 7m 16s mark.  During this period we can see extreme variations in Pitch, Roll, and Yaw that are not normal at all.  

In my opinion, what I see in the log file matches up very well with your original description of the behavior you saw.

These behaviors I see in the log file are definitely out of the ordinary and would probably cause me some concern about flying it again without investigating it further.

Unfortunately, I don't see anything in the log that explains why that behavior occurred.  Maybe someone else on here will be able to have a look.

I definitely would suggest checking out the props to see if they are damaged in any way.  I would try to manually spin the motors to feel if there is anything abnormal about the way they are spinning.

If you don't find anything conclusive, as others have mentioned, I would suggest opening a ticket with DJI support (http://dji.com/support).  I would recommend that you ask them to analyze your log file to see if they can determine what happened.  My guess is that they will also very likely ask you to make sure you are on the latest firmware version.  They will likely also recommend that you refresh the firmware to make sure that there are no problems with it.  They may also ask you to do an IMU calibration as well.  If you do this, make sure to do it on a perfectly level surface with the props removed.

Best of luck working though this.  Definitely keep us in the loop if you get any additional information.

KlooGee Posted at 2018-5-2 05:34
I've had a look at your log file and I believe the messages in red (Speed Error and GPS Position NoMatch) are very likely AFTER the fall when it hit your awning.

Looking at the log file, starting somewhere close to around the 6m 50s mark, we can see behavior that does not look normal at all.  Around that time frame (and it gets worse in the following seconds) we can see that the descent speed goes from just under 3m/s (which is the documented max descent speed for the Spark) to over 6m/s at around the 7m 16s mark.  During this period we can see extreme variations in Pitch, Roll, and Yaw that are not normal at all.  

Thanks very much for all your observations which are interesting. Obviously, you are much more familiar with the story the flight log tells as I looked at the numbers and did not see things the way you did, perhaps because I am unfamiliar with the norms.

There may have been a glitch in the electronics to shut down one or more motors and that would not show up in the flight log which shows on the physical movements (and battery condition). I accept that such things can happen - even sophisticated fighter jets will have unexpected technical problems but test pilots are trained to note everything happening even if they are about to crash!

My firmware was up to date and although I do not check the drone every time I fly, I did do a visual check on the items mentioned (as suggested by someone earlier) and the motors don't seem to have any resistance to being turned.

As I said, I have lost confidence in this unit even if it is repaired. It will take the fun out of flying when I keep wondering if it may happen again - and I may not be so lucky next time. It's like problems I've had with my cars - supposedly fixed and then suddenly, the problem comes back.

But I am keen to know what caused it to drop from the sky and hopefully, if the DJI people figure it out, they will be willing to share the information with me. If it is a problem they have never encountered and know they need to put in a fix, then I guess I have helped other Spark owners. There can be millions of conditions all over the world and there may be one which I had that was never expected. It's the same with trucks (I cover the auto industry) - there was one time a Ford truck broke its radiator fan and the engineers just could not figure out how it happened. It was only after many months that they figured out that it was a unique driving condition by some drivers in Malaysia that caused the whole engine to literally move and the blades hit it.

So I'll probably be using this as an excuse to get a Mavic Pro or Mavic Air. I considered getting another Spark but I am attracted to the transmission superiority of the Mavic Pro so maybe I will get it later in the month.

fans0c5f7811 Posted at 2018-5-2 05:50
Thanks very much for all your observations which are interesting. Obviously, you are much more familiar with the story the flight log tells as I looked at the numbers and did not see things the way you did, perhaps because I am unfamiliar with the norms.

There may have been a glitch in the electronics to shut down one or more motors and that would not show up in the flight log which shows on the physical movements (and battery condition). I accept that such things can happen - even sophisticated fighter jets will have unexpected technical problems but test pilots are trained to note everything happening even if they are about to crash!

Below is a screenshot from a small snippet (2 seconds) of your log file.  Each line represents one-tenth of a second in time difference.  VelocityX is the forward backward speed, VelocityY is the left to right speed, VelocityZ is the up and down speed.  Pitch is the angle the aircraft is from front to back, Roll is the angle the aircraft is from side to side, and Yaw is the direction the aircraft is pointed with 0 being north and 180/-180 being south.

FIrstly, if you just look at VelocityZ, you can see that it was descending at over 6m/s.  This is more than twice the document maximum descent speed of the Spark.  

Secondly, look at the values in the roll column.  At one point, within 1/10th of a second, the aircraft rolled from side to side from a positive 30 degrees to negative 40 degrees.  That is a 70 degree change in the side-to-side angle of the aircraft in 1/10th of a second. 2/10ths of a second before this, it swung from - 44.6 degrees to -2.7 degrees to 62 degrees.  This is a change of angle of about 100 degrees in 2/10ths of a second.  

Thirdly, look at the values in the pitch column.  You can see large swings from negative to positive angles.  

Fourthly, look at the Yaw values.  In this small time period of this screenshot, it has done a full circle.  Looking at the Yaw values for the full time period I mentioned in my previous post, you can see that it is essentially spinning for that entire period.  

Just looking at these values, it is quite easy to see that as it was descending rapidly, while spinning and oscillating wildly left/right and front/back.

I hope that you can get a response from DJI on what may have caused this, however, my guess is that they likely will not provide an explanation.  They will just let you know if they will replace it under warranty or not without providing an explanation.

Best of luck on whatever you decide to do moving forward!  Happy flying!

Another comprehensive analysis from KlooGee. I appreciate your efforts, while hoping I'll never need to ask a favour for the same service!

Also looking at the log, battery Cell 3 is always a bit lower voltage than the other two. It's not a very big difference, but perhaps it's enough to indicate that the cell isn't in perfect condition and isn't delivering full current. I'm interested in seeing how the results of your battery tests turn out. If it ends up being that battery, you can go back to flying worry-free.

KlooGee Posted at 2018-5-2 07:21
Below is a screenshot from a small snippet (2 seconds) of your log file.  Each line represents one-tenth of a second in time difference.  VelocityX is the forward backward speed, VelocityY is the left to right speed, VelocityZ is the up and down speed.  Pitch is the angle the aircraft is from front to back, Roll is the angle the aircraft is from side to side, and Yaw is the direction the aircraft is pointed with 0 being north and 180/-180 being south.

FIrstly, if you just look at VelocityZ, you can see that it was descending at over 6m/s.  This is more than twice the document maximum descent speed of the Spark.  

Wow! Those are really great insights. Yes, I saw it spinning as it fell and I also noticed the propellers windmilling.

I was going to chart the data but could get a clean conversion from CSV to Excel and didn't have the time to do it.

Thanks again for looking into my case.

JimFandango Posted at 2018-5-2 07:43
Another comprehensive analysis from KlooGee. I appreciate your efforts, while hoping I'll never need to ask a favour for the same service!

Thanks Jim!  I hope you (and everybody here) has multiple years ahead of safe flying and no need for analyzing of the logs!