So, it finally happened to me as well. I will put down all relevant data so that you may learn from my issue and finally DJI to really solve such a disturbing behaviour for a drone addressed to novice users.

1. No of flight: 106, no errors or missbeahaviour untill flight 106
2. Firmware versions: latest both drone and Remote
3. Apk version: DJI Go V 4.2.20, updated a day before the fly away
4. No usage of OTG, drone connected in 5.8 Ghz

Resume of the incident:

Fiirst flight of the day, one no real issue except of a lag in showing the distance from the home point. Beeing first usage of DJI Go V 4.2.20 I considered this related to the APK.

Second flight of the day, around 15 minutes later apk crashed 2 times before stabilizing when sarted, then record button did not worked. After cleaning up phone memory all went fine.

1. Minute 0:56 I noticed a tilting horizon while flying sideways to the right, the til was to the left which may occur if accelerometer is giving false information regarding the drone's angle. After I found this abnormal at minute 01:05 i started pulling back the drone which did but in a very large curve, describing half a circle(bowl effect?)
2. Minute 01:16, switched to atti mode with no audio warning.
3. Minute 01:17 magnetic field interference
4. Minute 01:23 weak GPS signal, position accuracy may be compromised
5. From here on I was fighting the drone which was not only drifting with the wind(this I am able to handle as I flyed several not GPS drones, hellis planes, whataver elese) but it was changing direction and speed drastically. After arriving in my area I tried to land with no response, then I switched to Sport mode and suddenly GPS was on again and the drone was behaving normally.
6. From minute 01:23 till minute 01:25 there is no info in the flight path altough the drone apperea 80 meters close to me that 2 second before.

My conclusions as following:

1. Tilted horizon indicates accelerometer issues and/or software issues
2. Magnetic field interference is a missinterpretation of accelerometer related issues.
3. Weak GPS signal has nothing to do with No'' of visible sattelites
4. While in atti mode the drone is not fully manual as it should and only drift in the wind but instead it's going wherever it wants, changing the speed and direction without input nor under the wind influence
5. Even so the drone is controlable but it requires a very good level of skills which is not something DJI is targeting with a selfie drone.


Last point which DJI NEEDS to address is to treat this issue like a safety point that it actually is, there is no coincidence now and a huge amount of users are/or will experience that in the near future.

The proposal of DJI on DJI support facebook page was to reset firmware and calibrate both compass and IMU, I am sorry guys but this MUST BE A JOKE!

I will specifically in a written form request for a complete refund for this product which represent a safety risk!

For sure, I am expecting a DJI official position!

Hello Dan,

Here we are hearing these complaints more than once a day now.

I think that there are SERIOUS ISSUES with the Android Software lately too. As you said most of the issues you experienced are Software Related.

Just also remember that when you have a Compass Error, it will then give you a GPS Error as they work hand in hand.

You also mentioned IMU Error, Tiltings and Toilet Bowl Effect. Did you check the Sensors - Compass / IMU before taking off ?


I had to spend another Euro750 to get an IPhone 7+ even though I have a Samsung S8+, to eliminate using Android.

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

The Spark is a very small drone, but if a New Version is to be produced, I suggested there is a Form of Redundancy, meaning that there will be 2 Compasses so that if 1 fails, you don't lose control of the Drone. There should be a work around also for the GPS and IMU too. I am not a Technical Engineer, and I think most People who are Amateurs don't need to do the thinking for DJI, as we are already paying good amount of money for the Drone.

As Pilots we don't need headaches to go into all this Details. Many are buying drones to take Photos and Videos and have Fun with there Family and Friends.

DJI have years of experience in Drones, they just need to tackle these Fly Away Issues, as nowadays the Software Updates are even making the Spark WORSE. People are losing Euro 600 everytime, and thats a Very Large Amount of Money.

Still I am in doubt this can be solved....

DMX_MT Posted at 2018-6-18 00:06
Hello Dan,

In case you are interested in the flight logs you may find them attached.Flight log

Dan O Posted at 2018-6-18 00:08
In case you are interested in the flight logs you may find them attached.Flight log

Thank you Dan ! Many will have a look tonight, in the morning many in Europe are working right now.

Have a Good Day Buddy !

DMX_MT Posted at 2018-6-18 00:19
Thank you Dan ! Many will have a look tonight, in the morning many in Europe are working right now.

Have a Good Day Buddy !

Yep,

I am in Europe as well, no IMU or compass issues were visible before flight, the values were green and very low.

I do understand that the drone may loose hardware and then switch to atti mode would still be something I could handle, but this was not the case as the drone was doing whatever it wanted to do. This is a pity for such a manufacturer and not only that it happened but mostly because they do not accept the fault and work on it.

regained GPS after switching to Sport. Interesting!

Is this your normal flight path or new to this location?

SparkChog Posted at 2018-6-18 01:06
regained GPS after switching to Sport. Interesting!

Is this your normal flight path or new to this location?

I am not sure it was related to switching to Sport, after I switched to sport I felt much more in control instantly. In flight log you may see speeds up to 72 KM/H, which is abnormal as well.

New flight path but in an area of max 40 km of my usual flight path.

Anyhow, never I fly in the same area as my interest is going to movie making.

Thanks for sharing your experience with us.

What was the wind speed at the time your flight ? Just curious to know as the drone's flight record shows lots of Yaw errors.

Ram-UK Posted at 2018-6-18 01:22
Thanks for sharing your experience with us.

What was the wind speed at the time your flight ? Just curious to know as the drone's flight record shows lots of Yaw errors.

The wind was close to zero on ground level, as the altitude was 24 meters I suspect no issues over there as well. No high wind messages and 15 min before have flown at 90 meters high in the same area with no issue at all.

I must insist that whatever external factor we may search it's time to accept that this is design issue, either software or hardware.

I have flown the P3P in the hardest conditions ever and never had any hard times with that thing.

This behavior the Spark have shown I used to experience with a cheap Chinese drone called Cherrson cx 20, that thing was toilet bowling every now and then and flew away 2 times, once to be found 1 week later and 3 km away.

Dan O Posted at 2018-6-18 00:08
In case you are interested in the flight logs you may find them attached.Flight log

Thank you Dan, most will comment about what may have caused the issue of YAW ERROR.

Dan O Posted at 2018-6-18 00:24
Yep,

I am in Europe as well, no IMU or compass issues were visible before flight, the values were green and very low.

So you checked the Compass / IMU before flight. Very well.

As Ram has told you it maybe me the Wind Gusts in Higher Altitudes or maybe you came across some Magnetic Interference maybe from Electrical Antennas / Wires.

You look very Technical and know your Drone very well. This is shown as you manage to Land Safely even though you have Trouble. Welldone on that Dan !

DMX_MT Posted at 2018-6-18 02:21
So you checked them before flight. Very well.

As Ram has told you it maybe me the Wind Gusts in Higher Altitudes or maybe you came across some Magnetic Interference maybe from Electrical Antennas / Wires.

Thank you,

It is never a concern to fly a drone in atti  or even in full manual as I have some flying experience.

there you go, my flight record in another site is showing more info. Funny thou that on sensor state there was no gps drop nor compass error, maximum turn rate was 10.

https://forum33.djicdn.com/data/attachment/forum/file:///C:/Users/doros/Desktop/Screenshot_20180617-203906.png

Dan, from the Flight Log it clearly shows that the - YAW ERROR is the cause.

You said that its something do with the IMU, the Yaw - Axis, right ?

Do you know what causes the YAW ERROR, please ?

Thank God you have experience and landed safely. Other Beginners would have lost the Spark.


Can DJI Support tell us what the 'YAW ERROR' is, and how it can be avoided please ??

Dan O Posted at 2018-6-18 02:45
Thank you,

It is never a concern to fly a drone in atti  or even in full manual as I have some flying experience.

I have PMed some DJI Moderator to come and explain what this YAW ERROR is.

Its very unsafe to know about an Error and dont know its cause and  don't have a solution how to avoid it.

We really need to address this Problem as its becoming too common !

Dan, until they answer have a look at my post - https://forum.dji.com/thread-151297-1-1.html

Maybe you find some New Information there.

I have attached 4 interesting videos.

DMX_MT Posted at 2018-6-18 03:40
Dan, until they answer have a look at my post - https://forum.dji.com/thread-151297-1-1.html

Maybe you find some New Information there.

Thank you once again,

I do not really understand the yaw error but I see it as follows:

Possibility one: between the compass and IMU there is a non coherence in therms of heading.
Possibility two: compass is reading that the drone is yawing at a higher speed than maximum speed allowed by the  IMU.

As the issue became higher in therms of erratic flight when changing the input direction from forward to backwards I tend to believe the issue comes from point one. Also this could explain why when flying to the right the drone was also drifting backwards as the accelerator could not understand the drone inclination.  Same explanation is plausible also for the tilted horizon considering that the gimbal keeps the horizon leveled based on the information coming from accelerometer.

One on top of the other DJI should/need to make it clear for us as soon as possible.

Hello. We apologize for the inconvenience that you're having right now. I just want to let you know that we already raised this concern to our engineers. We will keep you updated. Your patience will be much appreciated.

DJI Wanda Posted at 2018-6-18 04:29
Hello. We apologize for the inconvenience that you're having right now. I just want to let you know that we already raised this concern to our engineers. We will keep you updated. Your patience will be much appreciated.

Just to be clear DJI Wanda,

The engineers are aware of multiple users experiencing YAW Error issues?
I had the YAW errors once before with a unit and received a new one (not refurbished) from service and that seemed to resolve the issue.

DMX_MT Posted at 2018-6-18 03:40
Dan, until they answer have a look at my post - https://forum.dji.com/thread-151297-1-1.html

Maybe you find some New Information there.

Very good consolidation of data's that should be known of any drone user.

pmshop Posted at 2018-6-18 04:40
Just to be clear DJI Wanda,

The engineers are aware of multiple users experiencing YAW Error issues?

Thank you for your information,

It is indeed what I am expecting from DJI instead of never ending we will check...bla bla bla.  This errors are for sure not new and you are daily replacing drones for this so I see no reason to go to all this show over and over again.

On top of that, if no clear information and commitment is done by DJI that a new unit will not have the same issues I expect a potential refund to be agreed.

pmshop Posted at 2018-6-18 04:40
Just to be clear DJI Wanda,

The engineers are aware of multiple users experiencing YAW Error issues?

We cannot provide any information about it yet since we don't have any available troubleshooting for this kind of error. That's why we're going to confirm directly to our engineers if they already encountered this error and check if they have some solutions.

DJI Wanda Posted at 2018-6-18 04:51
We cannot provide any information about it yet since we don't have any available troubleshooting for this kind of error. That's why we're going to confirm directly to our engineers if they already encountered this error and check if they have some solutions.

I appreciate!

The engineers should have my logs already from when I had the issue back in April '18

pmshop Posted at 2018-6-18 06:22
I appreciate!

The engineers should have my logs already from when I had the issue back in April '18

They do. It will be saved in their system once you synchronized it on your app.

Dan O Posted at 2018-6-18 04:20
Thank you once again,

I do not really understand the yaw error but I see it as follows:

Thank you Dan,

Both of your possibilities are possible and you went in so much details, I really am enjoying talking to such a Tech Guy like you Dan !

I just knew about the Possibility One infact let me attach a Video where the Compass and IMU wont match the heading.

The Second Possibility is New to me. Thank you for pointing it out ! I wish that DJI or someone can confirm from where this is coming.

DJI Wanda Posted at 2018-6-18 04:51
We cannot provide any information about it yet since we don't have any available troubleshooting for this kind of error. That's why we're going to confirm directly to our engineers if they already encountered this error and check if they have some solutions.

Dear DJI Wanda,

This YAW Error is happening almost daily, we as Spark Pilot would like to know what this Error Is, and we have spent over $500 dollars on a Drone, and usually the Possible Errors are explained at the back of each Manual. Every Pilot has the right to know how to avoid Dangerous Situations, and this Yaw Error is creating a lot of Fly Aways and Pilots losing their Drones.

We will appreciate the Help of the DJI Technicians. Please threat as URGENT as this needs to be solved.

Thank you so much.

Dan O Posted at 2018-6-18 04:49
Thank you for your information,

It is indeed what I am expecting from DJI instead of never ending we will check...bla bla bla.  This errors are for sure not new and you are daily replacing drones for this so I see no reason to go to all this show over and over again.

Totally agree with you Dan. If I knew that Spark will be such a problem I wouldn't have bought it at first place and would have looked somewhere else.

DJI Support must address these Errors please, if they want to keep their Customers Happy.

DMX_MT Posted at 2018-6-18 07:04
Dear DJI Wanda,

This YAW Error is happening almost daily, we as Spark Pilot would like to know what this Error Is, and we have spent over $500 dollars on a Drone, and usually the Possible Errors are explained at the back of each Manual. Every Pilot has the right to know how to avoid Dangerous Situations, and this Yaw Error is creating a lot of Fly Aways and Pilots losing their Drones.

I understand how much this is important to you all. We already raised this up to our engineers. We will keep you updated. Thank you for your patience.

pmshop Posted at 2018-6-18 04:40
Just to be clear DJI Wanda,

The engineers are aware of multiple users experiencing YAW Error issues?

Hey Pmshop,

But what if the New One (Hope Not) will start giving the same Yaw Error, do you know from where this issue came from ?

How can you solve it when DJI Care Refresh or the Guarantee are expired ? Will you pay DJI for this UnExplained Yaw Error ?

Nice to see you here Pmshop !

DJI Wanda Posted at 2018-6-18 07:06
I understand how much this is important to you all. We already raised this up to our engineers. We will keep you updated. Thank you for your patience.

Thank you so much Dear DJI Wanda !

I wish we solve this YAW ERROR.

DMX_MT Posted at 2018-6-18 07:09
Thank you so much Dear DJI Wanda !

I wish we solve this YAW ERROR.

My pleasure. We will do our best to help you with this. Thank you all for your patience as well. We will update you once we get an answer from our repair team.

Latest firmware update is UNSTABLE on Android. Choppy video, record button sticks. Unreliable. It's like DJI just cares if Apple products interface well.
Got an Android? You're out of luck.

JohnnySuperDad Posted at 2018-6-18 10:17
Latest firmware update is UNSTABLE on Android. Choppy video, record button sticks. Unreliable. It's like DJI just cares if Apple products interface well.
Got an Android? You're out of luck.

I agree with you Johnny !



Many have suggested that I switch to IOS.

DMX_MT Posted at 2018-6-18 10:18
I agree with you Johnny !

I'll switch drone companies before I switch phones.

Dan,

DMX_MT pointed me to you post – thanks DMX_MT since this case is fairly fascinating!

I think I know what happened – read on:

0m 56 sec – tilted horizon.
Sparks roll is consistent with your sudden full speed to the right yank. Sparks gimbal simply did not manage to compensate. While it happens often in Sports mode during sudden turns, it may, sporadically happen when you yank full speed to the side as you did. I cannot say for sure without seeing video but my bet is on gimbal motor having trouble compensating your sudden move. Sparks gimbal motor is not that strong.

Your flight path between sec 56 and 1m 4 sec is all wrong. It seems to be a large arc without your input of steering it in the arc. This points to either fairly strong wind from NW or disagreement between compass / IMU and GPS position. Your commanded position is slightly different than physical position and Spark compensates. However, looking at Sparks pitch I do not recognize strong wind – perhaps a bit of light breeze. You also mention there was no significant wind which confirms what I see in the logs. My bet is on beginning of toilet bowl effect which Spark does not recognize yet and sticks to P-GPS mode.

At 1m 4 secs you start to fly backwards with Spark facing NNW and your very small input to the right.
Flight paths seems to draw the same counter clockwise curved line. Here, the results of phantom maps and google maps positioning do not quite match so I am not 100% sure but it seems continuation of the pattern which started at 56 seconds. Again – my bet goes towards toilet bowl effect which is not yet recognized by Spark.

At 1 m 15 seconds Sparks finally realizes that there is a mismatch and one second later drops into ATTI (correct behaviour which may have saved your drone ).

At 1 m 21 sec it gives GPS warning with 18 satellites. This is unusual but not necessarily wrong. Spatial separation of satellites could be narrow giving Spark troubles in calculating its position. Again – this would be unusual. On top of that, Spark continues to keep track of its Geo position. I bet on Sparks software getting confused with small inconsistencies and errors between three key components – compass, IMU and GPS positioning.

There is an interesting bit happening at 1 m 23 sec. As soon as you put Spark in Sport mode, yaw errors disappear. Speed change and disappearance of yaw errors happen 0.3 sec before logs register Sport mode which proves to me again that some discrepancies between messages and metrics logs do exist in Spark. I have seen it countless times. Not a big deal once you understand this.

The last 23 seconds are uneventful. You bring it back and land.

I do not see anything in the vicinity which would justify compass errors. You fly high, there are some power lines close to the road but nothing unusual. Not something which should mess with your compass. The problems most likely started with your sudden direction change to the right at 55.4 sec. Perhaps compass and IMU misalignment which caused Spark to drop to ATTI later. The problems ended with your sudden easing on sticks at 1 m 21.8 s. While Yaw error persisted in logs for another second, I believe it was through inertia before s/w figured out that things got fixed.

My conclusion – it is an issue with Sparks IMU and its gyroscope. Something got stuck with sudden move and dislodged with another sudden move. This would cause yaw errors as a result of misalignment of correct compass readings and stuck gyroscope. IMU in Spark is fragile. I would certainly request DJI to do similar detailed analysis and, if they conclude that this is not IMU gyroscope issue, they should tell you what it is than. I don’t believe for a second that it was due to un-calibrated compass or un-calibrated IMU. Too much of a co-incident when the problem appeared and disappeared.

Cheers!

Is it possible for DJI programmers to decrease instability of the drone, caused by conflict between Compass and GPS (started with multiple YAW errors and resolved nowadays by switching off GPS and entering ATTI mode), by using the hardware of smartphone - its GPS and compass, which most likely are of much better quality and sensitivity, than built in AC

DMX_MT Posted at 2018-6-18 07:08
Hey Pmshop,

But what if the New One (Hope Not) will start giving the same Yaw Error, do you know from where this issue came from ?

1) Mine was covered under warranty and

2) fortunately for me I saw the DJI Care Refresh + offer

Now Sparky 1 has 4 replacements left and Sparky 2 has 5 replacements left

JohnnySuperDad Posted at 2018-6-18 10:17
Latest firmware update is UNSTABLE on Android. Choppy video, record button sticks. Unreliable. It's like DJI just cares if Apple products interface well.
Got an Android? You're out of luck.

It is the other way around.
Firmware has been good since 0900
It is the Android app since 4.2.15 that has been unstable.
Multiple forum comments on rolling back to 4.2.15 Android.

But App, firmware or choppy video is not what we are discussing here...

Specifically, an ongoing YAW, compass IMU error

JohnnySuperDad Posted at 2018-6-18 10:17
Latest firmware update is UNSTABLE on Android. Choppy video, record button sticks. Unreliable. It's like DJI just cares if Apple products interface well.
Got an Android? You're out of luck.

Got the same problems and been switching between Android phones to try and pinpoint the problem including installing new SD cards etc.  

pmshop Posted at 2018-6-18 11:41
It is the other way around.
Firmware has been good since 0900
It is the Android app since 4.2.15 that has been unstable.

pmshop,

You are correct - let's put aside firmware or DJI GO s/w issues here.

I am fairly certain that what we are discussing here is IMU mechanical problem were some part, most likely gyroscope, got stuck with sudden movement and dislodged again with another sudden movement. See my post and explanation above in this thread.
I challenge DJI to prove me wrong :-).

Dan may not see this problem now (IMU is working again) but it can reappear any time.
Spark's IMU seems fairly fragile and this is why we see a good number of yaw errors. Sometimes it is magnetic interference but not this time - this time it is h/w issue.  

Again - DJI, please, please prove me wrong :-). I'll fly to China to buy you coffee if you can do that :-)

pmshop Posted at 2018-6-18 11:36
1) Mine was covered under warranty and

2) fortunately for me I saw the DJI Care Refresh + offer

Wohooooo ! Welldone PM ! Still Fly Safe !