I was flying from the top of a mountain, with a fresh compass calibration, no metal/spare-battery or controller nearby.

In mid air it makes a wierd right turn and goes ATTI. I barelt managed to regain control, fly home and land safely.

This is now my 2nd time ATTI with the spark. The first time I crashed and was issued a free a repair from DJI.

I'm disappointed whith the spark as it might fly away at any moment. I did the proper compass calibration and i checked that the IMU was good before takeoff.

Any thoughs on what is causing this?

And what do I do? Does Mavic Air and Mavic Pro have the same problem?

-Regards Erlend

Here is the log:
http://www.phantomhelp.com/logviewer/8FF858NU3Q8Y4X7LPNWD/

Original Log file:
https://drive.google.com/open?id ... MSu0ZvzeEixLPCwT-gt

Video: (only plays in VLC because it has some error)
https://drive.google.com/open?id ... 9gjcikPX-bNEoqwkcKq

Hi, thanks for reaching DJI Forum. We do feel sorry to hear that you are having issues with your DJI Spark. May we ask for the firmware versions of your aircraft and the remote controller?

Erlend,

Too many errors and too dynamic piloting to figure out exactly what was happening.
Your logs show total confusion of Spark's navigational systems.

This is what I would do now:
- calibrate IMU. It does not matter if DJI Go says it is normal or not - just do it.
- calibrate compass again but this time do it very slowly (rotate your Spark slowly during calibration). You may want to rotate twice (720 degrees) on both axis.

I will not repeat how important it is to do calibration far away from any potential interference. This does include rocks (I have noticed you are flying in rocky terrain). If rocks contain hematite or magnetite you may be in trouble.

IMU calibration must be done very carefully as well. Level surface, no vibrations etc.

After you have done that, go out and fly. Start from palm launch - just in case rocky ground contains some magnetic or iron ore. Watch for any errors appearing on DJI GO. If they do - land. After several flights analyse your flight records for signs of speed, compass, yaw or GPS issues. If you see any again and you followed my steps with IMU and compass calibration, open ticket with DJI.

Good luck and let us know.

Mirek

This is normal after an update . You must take off with the SPARK slowly after an initial warm up to check if all the components are functional.

Mirek6 Posted at 2018-7-23 09:37
Erlend,

Too many errors and too dynamic piloting to figure out exactly what was happening.

Mountains are places for limestone rocks which cause heavy EMI interference to compass.

Hi guys, I appreciate your feedback.

I'm going to add a few more details about the circumstances leading up to the Spark going ATTI mode:
* Just 5 minutes before the flight when it went ATTI, I had a 15 minutes flight on my first battery. So there was no update between the first flight and the ATTI flight
* Calibrated the compass before the first flight, and everything was normal.
* The mountain is Liadalsnipa, and the rocktype is granit quartz diorite. It is grey and has no iron, pyrite or magnetic properties.

Here is a video I made filming the compass deviation immediatly after the ATTI flight. It shows values are not super low, but within the limits.
https://photos.app.goo.gl/243JpSckHeAbv2BU9

Here are the flights i made that day on the mountain:
1) Calibrated before takeoff, no problems
http://www.phantomhelp.com/logviewer/TDJ4Q6RQOOCUPVBHALZ8/
2) No problem
http://www.phantomhelp.com/logviewer/SAA5O3IRNM4VNT2H8IR9/
3) ATTI, almost fly-away
http://www.phantomhelp.com/logviewer/8FF858NU3Q8Y4X7LPNWD/
4) Recalibrated compass again and did a test flight, no problems
http://www.phantomhelp.com/logviewer/OOWZLH5KJZQPK7PB5WD3/

How would you approach this differently to avoid ATTI?

-Erlend

DJI Gamora Posted at 2018-7-22 14:13
Hi, thanks for reaching DJI Forum. We do feel sorry to hear that you are having issues with your DJI Spark. May we ask for the firmware versions of your aircraft and the remote controller?

[view_image]

Hi
Aircraft Firmware: V 01.00.0900
Remote Controller Firmware: V 01.00.0400

I believe this was the most recent update  up until the 19.07.2018 release?

Regards
-Erlend

Erlend Rongen Posted at 2018-7-23 13:08
Hi guys, I appreciate your feedback.

I'm going to add a few more details about the circumstances leading up to the Spark going ATTI mode:

Why did you calibrate the compass? Did the DJIGo app ask for this?
If not requested then you do not have to calibrate it because a bad calibration (you can not detect magnetic interference) can distract the well-functioning compass.
Do not fix what works!

If you've been flying a lot without a mistake then the compass is well calibrated, it can only be spoiled ...

Probably the new firmware is trying to fix this weakness because it is the only one that is writing, support for OTG (which worked fine).

Hi. Thank you for the additional information. I believe your aircraft is outdated already. Please try to update the firmware of your aircraft to avoid errors and malfunctioning. The aircraft's firmware should be at .1000 and the remote should be at .0600 version.

Erlend Rongen Posted at 2018-7-23 13:40
Hi
Aircraft Firmware: V 01.00.0900
Remote Controller Firmware: V 01.00.0400

That was the December 2017 release.
DJI Stephen is right

Update your firmware and try again.

However, I think you got a bad replacement unit.
There is one error I noticed that was extremely troubling considering the opposite fact.
Weak GPS Signal yet you had 23 sats locked???
Doesn't look good.

And, there has been a rash of YAW Errors going around.

I would say go ahead and try again after updating.
But if it is still screwy, open a service request and get it back to them ASAP.

By the way (and it doesn't mean much) did you get a refurbished craft in exchange?
You can tell by the serial number.
"R" in the serial = refurb
no "R" in the serial = new

pmshop Posted at 2018-7-23 14:05
That was the December 2017 release.
DJI Stephen is right

Hi

Thanks for your reply. The drone threw the GPS error on the previous fly-away as well. I think this happens when the navigation system gets confused and does not know weather to trust the compass or the GPS. I'll see if I can find the log file from the crash i had earlier.

DJI Stephen Posted at 2018-7-23 13:51
Hi. Thank you for the additional information. I believe your aircraft is outdated already. Please try to update the firmware of your aircraft to avoid errors and malfunctioning. The aircraft's firmware should be at .1000 and the remote should be at .0600 version.

Hi thanks for your reply

Yes. But I could not update the firmware before it was released. My flight was 2018.07.09 and the firmware was released 2018.07.19. My flight was on the latest available firmware.

-Erlend

Erlend Rongen Posted at 2018-7-23 14:31
Hi

Thanks for your reply. The drone threw the GPS error on the previous fly-away as well. I think this happens when the navigation system gets confused and does not know weather to trust the compass or the GPS. I'll see if I can find the log file from the crash i had earlier.

Here is the log from the first time it went ATTI.

The drone did not throw a GPS error.

http://www.phantomhelp.com/logviewer/MAR7I40SHMLWHTKJ2J8B/

(the ships was ofcourse not there when i was flying)

pmshop Posted at 2018-7-23 14:05
That was the December 2017 release.
DJI Stephen is right

pmshop,

This R in serial number is a myth. I bought a new Spark from the store and it did have R in serial number.
I got refurbished one from DJI and it also had R.

Yes - DJI told me that R in serial number means that the unit is refurbished but how does this explain an R in my new one?

I am not sure we can use this argument any more.

Mirek

Erlend Rongen Posted at 2018-7-23 13:28
Here are the flights i made that day on the mountain:
1) Calibrated before takeoff, no problems
http://www.phantomhelp.com/logviewer/TDJ4Q6RQOOCUPVBHALZ8/

Erlend,

I will address your questions and several answers which were posted afterwards.

• I believe that quartz rocks do not have any EMI properties – you are right. So this is not the issue.
• You are also right in your response to pmshop. Very often when there are multiple errors (compass, yaw, speed, GPS) it means that Spark systems are confused. The source problem could be one (for example compass) but it often manifests itself as multiple warnings. I am confident that your GPS unit is just fine.
• Jakab is mistaken. The fact that DJI GO shows everything normal does not mean that everything is normal. DJI GO is buggy. Yes, bad calibration can cause more damage than no calibration but I do trust that you know very well how to calibrate. First thing to do when any such abnormalities start happening and there is no apparent reason for them (no interference) is to re-calibrate compass and IMU whether or not DJI GO prompts you to do it. In my experience, it almost always helps. Of course there are rare circumstances where Spark’s h/w is at fault.
• S.J. Is right – do 30 second warm up before you start to fly more aggressively.
• Your firmware is just fine. Yes, it is always a good idea to upgrade to new firmware but I never do that within first month of firmware release. Let others do the testing and see if it really works or not. We have no idea what DJI included in newest firmware outside of OTG support for Android. They did not tell us. I assume that there are some fixes to ATTI issues and fly-away problems but I am not sure. Having said that I am not going to advise against upgrading to new firmware. New firmware is, most likely, better than last 0900 version which you have
  

You are asking: “How would you approach this differently to avoid ATTI?”

If you followed all the steps in careful calibration and if you do not fly in area with magnetic or electromagnetic disturbances than you did most you can. I would also pay attention where you store your Spark. For example if you drive to the flight site with Spark sitting close to the subwoofer in your car, you may be up for an unpleasant surprise. Proximity to magnetic materials or electromagnetic fields may de-calibrate your compass. Be vigilant – if problems reoccur after careful calibration, the chances are that you have faulty unit. Again – if you see errors – calibrate both compass and IMU.

One more tip. If, during flight, your Spark drops to ATTI, try to switch to Sport mode. We suspect (based on experience) that Sport mode desensitizes some of the sensory input and is more forgiving to yaw and compass errors. This is done to prevent errors during aggressive flying and turns. Switching to Sport did save many a Spark but does not always work.

Mirek

Mirek6 Posted at 2018-7-24 05:28
pmshop,

This R in serial number is a myth. I bought a new Spark from the store and it did have R in serial number.

That is what came straight from DJI "R" in serial is refurb
Only conclusion is you got a refurb sold as new - wouldn't be the first time on any product.
Remember Amazon shipped out a bunch of returns as new?

pmshop Posted at 2018-7-24 05:45
That is what came straight from DJI "R" in serial is refurb
Only conclusion is you got a refurb sold as new - wouldn't be the first time on any product.
Remember Amazon shipped out a bunch of returns as new?

Agreed potentially a refurbished boxed as new. It happens

S.J Posted at 2018-7-23 11:08
Mountains are places for limestone rocks which cause heavy EMI interference to compass.

Limestone does not have any magnetic properties. And those mountains are not limestone.

.900 firmware, no surprise there(!)

pmshop Posted at 2018-7-24 05:45
That is what came straight from DJI "R" in serial is refurb
Only conclusion is you got a refurb sold as new - wouldn't be the first time on any product.
Remember Amazon shipped out a bunch of returns as new?

Anything is possible. I got a new bundle from Amazon. My new Spark promptly developed gimbal issues and was replaced by refurbished one by DJI. No issues with refurbished so far - fingers crossed.

I will just refrain from advising people that "R" in serial number means refurbished. My experience shows that it may not be true. But, agreed, it does not definitely say that it is not true given your Amazon example. Don't know.

Mirek

Mirek6 Posted at 2018-7-24 05:33
Erlend,

I will address your questions and several answers which were posted afterwards.

HI

Thanks Mirek6 for a very comprehensive answer and well founded input in preventing ATTI.

I really like the 30 first second test-run with forward-backward, left-rigt, and 360 yaw can give good clues about the state of the drones navigation system. In fact the drone seemed a bit wierd in the beginning of the flight. I checked the tracklog against control inputs and look what I found!

With only backward stick input, the drone was not traveling in a straight line.

I guess if the drone is not travelling in straigt lines during the first 30 seconds, then LAND safely as soon as possible.

BTW, is there a way of sharing flights from the flight list in the DJI Go app, so other users can make up their mind about what happens?

-Erlend

I'm sorry for your troubles with your Spark!  I'm happy that you were able to fly it back and prevent a fly-away!  But - I watched your video you linked in the 1st post and - how beautiful!!  What am amazing view and climb/hike.  You got some pretty cool shots

Bender1031 Posted at 2018-7-24 07:30
I'm sorry for your troubles with your Spark!  I'm happy that you were able to fly it back and prevent a fly-away!  But - I watched your video you linked in the 1st post and - how beautiful!!  What am amazing view and climb/hike.  You got some pretty cool shots

Thanks Yes it is a pretty exposed peak. Luckily we did not have to walk down the steep path. The wind finally allowed us to fly down with paragliders

Erlend Rongen Posted at 2018-7-24 07:54
Thanks Yes it is a pretty exposed peak. Luckily we did not have to walk down the steep path. The wind finally allowed us to fly down with paragliders

Paragliders!!  Wow - how fun!  Were you going to record any of that flight with your Spark before it went crazy?

Bender1031 Posted at 2018-7-24 08:08
Paragliders!!  Wow - how fun!  Were you going to record any of that flight with your Spark before it went crazy?

I had a helmet cam on the flight down from the mountain. It would look awsome whith drone video from takeoff and chase into the air.

Erlend Rongen Posted at 2018-7-24 08:19
I had a helmet cam on the flight down from the mountain. It would look awsome whith drone video from takeoff and chase into the air.

Cool!  Do you have that helmet video to share?

It'd be awesome with the drone!  But wouldn't you have to stay behind to land the drone?

Erlend Rongen Posted at 2018-7-24 07:11
HI

Thanks Mirek6 for a very comprehensive answer and well founded input in preventing ATTI.

Erlend,

Your backward flight path looks like trouble.
I have seen it many times and, inevitably (in all the cases I have seen), it meant that there is a discrepancy between Spark compass direction and Spark’s physical direction.

Here is how it works.
With well calibrated compass Spark’s direction and compass direction are in line. When you move your stick, Spark’s firmware calculates so called COMMANDED position - the position you ask your Spark to be in in a next while (after X ms). The commanded position is calculated based on where the compass is showing the Spark is facing, where the Spark is now (its exact GPS position) and the vector of your stick movement. The commanded position is simply a GPS position where the Spark should be in the next few moments (after X ms).

Spark starts flying. After X ms passed Spark checks its physical position. If it is equal to commanded position all is fine. It calculates new commanded position etc. If, however, physical position is not the same as commanded position Spark calculates adjustment vector which it adds to the directional vector from your stick input. If compass misalignment is constant, the flight path will be superimposed with a series of adjustments and will start to look like an arc. Read about toilet bowl effect and you will understand what I am talking about. And here is a rub. This ellipsoidal path is consistent with toilet bowl effect which is quite dangerous phenomenon (difficult to control if you are in GPS mode). So DJI design is to drop Spark to ATTI to allow full manual control. In most cases I have seen so far, the ellipsoidal path was sooner or later followed by drop to ATTI.

There is another factor – wind. If the wind was blowing from your right side to the left, it could have caused ellipsoidal path. But … I do not see arc in white line. Also, I only saw ellipsoidal path consistent with Spark fighting the wind when its sensor systems were already confused (in other cases I analyzed). And it also, inevitably, lead to later ATTI.

In the analysis of other similar cases (wind or no wind), DJI ended up replacing drone if it was still on warranty. This indicates issue with their h/w or firmware. We speculated that the December 2017 firmware (0900) is weak and can misinterpret sensory input causing constant adjustments and drop to ATTI. It is speculation, because, in a vast majority of Sparks (including mine which is on 0900) you do not see this problem. Also, in a majority of cases where the Spark was recovered, calibration of IMU and compass helped.

You calibrated IMU and you calibrated compass. You still see the problem. Next step would be to upgrade to newest firmware. But before you do this, read posts in Software section of this forum. You will see what people think about newest firmware and what kind of trouble they experienced. Be aware of that.

If you can, upload this particular flight record to phantom site – I would like to see logs so my recommendation is not biased.

No, you cannot share your flights directly from you flight list in DJI Go App. If you want to share it with forum members, upload to phantom site and share the link with us.

Mirek

My Spark Flew away after 11 seconds of flight in a completely unobstructed and remote place.  At 500 feet away the Drone dropped reception to the controller and just flew away at 35 MPH never to be seen again. It could have flew into a highway traffic, kids in a park, path of an airplane. DJI looked at the flight data and said because it was in ATTI mode it was totally my fault. I feel this is ridiculous. They refuse to answer these questions:
1) Even thought it was in ATTI why did it loose connection to my controller in a remote area with no RF interference or physical obstructions?
2) There are audible warnings about taking off, landing, low battery, etc. but if you are at such a risk of  losing your Drone when unknowing it is in ATTI mode, why is there not a similar warning?
3) The area I was at had STRONG GPS signal which I tested, why will they not answer my question why the failure of the GPS was not the true problem. Search the GPS issues with Spark and see how many have had iGPS issues!
4) When the Drone lost connectivity to the controller, why did it not stop and hover as they said it should have done? I could have retrieved it had this happened.

I believe DJI should replace my Drone for 3 reasons:
a) GPS failure
b) Drone did not stop and hover when losing connectivity (thus could result in very dangerous consequences). Two DJI agents said it should have stopped by design.
c)  Inadequate warnings when flying in a dangerous condition they call ATTI mode

I look forward to DJI answering these points

I had the same thing happen to me. Im not afraid of ATTI mode as i have used gyro stabilsed drones for years, and can handle drift.

The first spark i had combined compass errors to produce ATTI mode, during a return to home (i was only 10m away from home point and at about 80m altitude). the spark not only went into atti mode, it also yaw'd at full speed WITH the wind, my flight log shows it reached 74km/h in about 2 seconds.

I freaked out. This was only my second flight at over 30m. Those 2 seconds were enough to lose visual on the drone. Luckily i was able to return the drone manually in ATTI mode, without heart failure.

DJI needs to allow the spark to switch between ATTI mode and gps, to allow users to experience flight dynamics in ATTI mode, i feel this would allow us to practice emergency situations before they happen.

It turns out that first drone had a faulty compass and was replaced. Im both excited and hesitant to fly my new replacement spark. I really hope it doesnt let me down.