I hadn't much chance of regularly playing with Spark recently, but I take off couple of times a month.
I always - and I mean always - have to be prepared for rescueing the craft from ATTI mode.

After a few minutes of stable flight it starts reporting compass/speed/yaw errors and enters ATTI in some cases for up to 2 minutes. That makes it a nervous dance to take it down somehow. All that with 16-20 sats locked on cloudless calm days, though it tends to be a bit windy up above.
This is sometimes coupled with magnetic field interference messages, which I find hard to explain when the bird is 150-200m high, and there are no buildings, wires/cables, roads/rails... just nature all around.

This wasn't a regular behaviour last year, so I wonder what has changed?
I did compass calibration a few times, calibrated MRU as well, and the app doesn't report anything irregular before flights.

Any idea whats wrong? If I have to put up with this new feature, I am going to consider selling it unless I loose it first.

You could try to re-flash the firmware.
Or change to an older one, for a try.
The GPS will not be in use, as soon your bird gets into atti.
Compass gone, there ist just one, GPS is considered unverified, cause the bird does not know where north is.

Have you been in Sport, when it was happening?
-" That makes it a nervous dance to take it down somehow."-
Every stick input in atti is directly going to the bird, same like in Sport.

In the 2 minutes of atti, have you tried to "let it go" for a couple of seconds?

I had just a few atti "failures", but the first I try, now, is let the control go for some seconds (if in real LOS or with life feed in time) and just use the throttle for reaching maximum height to get over obstacles.
If possible!
And it seems to get back in P-GPS easier, this way.

I have the same problem: https://forum.dji.com/forum.php?mod=viewthread&tid=186437

I've contacted DJI and they said that my drone should be checked in DJI serice center.  The problem is that I bought my drone in US, but i live in Europe, so all the shipment cost should be one me.

Anyone have an idea what DJI could replace/check during examintaion at serivice shop? If this is simple IMU/compas calibraion, then I can do it by myself.

Its interesting that some Sparks experience this issue regularly, while others like mine do that only seldom.
Pure luck I guess....
What Spark firmware version are you using? Did you use a R/C?

Cheers,
Wolferl

I have the same issue, and the 'letting it go' advice could be worth a shot. Like you, I am now expecting these ATTI switches, so at least do not panic - I suspect some of these fly aways are related to the panic that
sets in when a pilot first gets the ATTI shock...my first 'fly away' was certainly that!

S-e-ven Posted at 4-23 03:20
You could try to re-flash the firmware.
Or change to an older one, for a try.
The GPS will not be in use, as soon your bird gets into atti.

I didn't use Sport mode lately, but I did put on full throttle with disabled vision sensors to fly some hundreds of meters away. Is it the speed that makes the compass unreliable?

When there wasn't a scary wind up there, yes I did let it loose for some seconds, but that didn't seem to help. At least not the last time when it stayed in ATTI for 2 minutes.

Also ATTI is not the same as Sport, since in Sport it does hold it's position. In ATTI it just goes in whatever direction it feels like, so there's not much choice but to stare into the sun the try to control it the opposite direction.

Wolferl Posted at 4-23 03:44
Its interesting that some Sparks experience this issue regularly, while others like mine do that only seldom.
Pure luck I guess....
What Spark firmware version are you using? Did you use a R/C?

Latest (but that's quite old as I recall) official firmware on all devices, with RC, no otg-cable. Latest app..., nothing out of ordinary.

munchausen Posted at 4-23 05:10
I didn't use Sport mode lately, but I did put on full throttle with disabled vision sensors to fly some hundreds of meters away. Is it the speed that makes the compass unreliable?

When there wasn't a scary wind up there, yes I did let it loose for some seconds, but that didn't seem to help. At least not the last time when it stayed in ATTI for 2 minutes.

Atti is like Sport without GPS
Full reactions on stick input!
Which I think, are the reason for not "calming down", if in atti!
Often the Compass ist "shaked". That's not getting better, with maneuvering hard to control the bird.
This is the reason that I try to give it some time without for-/back-/sideward inputs
In case that seems possible.
The last time it worked out in 5 seconds, about.
But perhaps my birds are not that affected, more stabil, as yours/others.
Dji could know,  but can we?

I’m pretty sure the DJI Techs have an idea why these issues manifest themselves but are keeping quiet as a public admission could result in costly settlements. Just one conspiracy theory ;-)

I must admit I do have a thought it could be a design issue. I use a number of different flight controllers, these include the Naza/Wookong/A2 made by DJI and opensource APM/Pixhawk. All of these flight controllers use a GPS unit with its own built in compass/magnetometer, the Pixhawk also has a second compass’s mounted on the flight controller board as a backup. The Spark is not as refined as the compass is separate from both the flight controller and the GPS module, the GPS is at the rear of the Spark and the compass is at the front. This opens up the possibility for a momentary break in the electrical connection between the GPS/IMU modules.
Regards

Nidge

Seems like a spate of these issues lately.  I had one this past weekend.  Lots of sats, did a Helix and boom, Atti mode.  This was on my own property, clear area, all natural, flown over it at least 80 times.

I think, that the Spark is adding disturbances, till it gets restarted.
Meaning:
In case, I get a warning "calibrate compass or move Spark", when trying to start from something, and moving it away clears that message, there was some metal in the "something"
So it seems good, but: I think, now it just needs a little other "something" , mid air, to get a compass failure, a yaw-/speed error, Sport in windy conditions with quick(er) direction changes, whatever.
Since the bird gotten already "something", it adds up and ends quicker in atti.
Perhaps that is rubbish, but since I restart my Spark after every warning, I seem to have even less "magnetic ...." in the air.
Also I care a bit more the wind warnings, tend to fly less in sport, for quicker speed often diagonal, instead.
It is not 50 kph, but still 30 then ;-)
And less inclination for the Spark!

S-e-ven Posted at 4-23 10:02
I think, that the Spark is adding disturbances, till it gets restarted.
Meaning:
In case, I get a warning "calibrate compass or move Spark", when trying to start from something, and moving it away clears that message, there was some metal in the "something"

I don’t think your comments are “rubbish”, S-e-ven, as without any official comment from DJI all we as customers can do is speculate based on our own experiences.

I don’t know whether my experience is more luck than management but to date I’ve not experienced any of these issues.... yet.

One thing I have done is program my Sport switch as an ATTI switch, similar method that Wolferi has detailed in this forum, for those times I want to speed things up a bit. If I ever do experience a forced ATTI mode while flying in P-GPS mode it will be interesting to see if I can manually force it back into P-GPS by flicking the mode switch back and forth.

Regards

Nidge.

Wolferl Posted at 4-23 03:44
Its interesting that some Sparks experience this issue regularly, while others like mine do that only seldom.
Pure luck I guess....
What Spark firmware version are you using? Did you use a R/C?

I use the latest firmware and I fly always with RC.
I fully agree with you that it is very strange that some people seems to use spark without those issues...
I'm very curious what DJI would fix in my spark if I would send it for repair in US (and if it would be permanent fix)...

Nidge Posted at 4-23 11:01
I don’t think your comments are “rubbish”, S-e-ven, as without any official comment from DJI all we as customers can do is speculate based on our own experiences.

I don’t know whether my experience is more luck than management but to date I’ve not experienced any of these issues.... yet.

You can't switch a bird, that is in real Atti, back to P-GPS. You just can switch a P-GPS bird to a mode, where it works like in atti.
It still is in GPS mode, has compass, no interferences, but is ignoring this, when the switch is used.
Aside, if using the atti switch to fly around like in sport (speed things up), you risk loosing the bird, in case you lose the RC connection. I do not expect, it switches the forced atti back, then.

I just had two flights in a row where I received a yaw error/magnetic field interference, then switched to ATTI.    First flight was in sport mode, full stick forward, several hundred feet up and it occured.  2nd flight I was ascending while moving forward and it occurred at 600 ft.  Compass was calibrated.
**edit**
  According to the flight log, It looks like the moment it switched to ATTI, when I was trying to get back to home, I was applying forward pitch, but it is showing (incorrectly) as rolling right, so the compass error is confirmed.

I don't see how there is interference up that high.

I would check on the micro SD card first.

When I first flew my Spark, I always experienced switches to ATTI mode. And then I noticed that the issue's gone when I only flew around and didn't record anything. After I replace the micro SD card with the fastest I could find (Sandisk Extreme Pro), my Spark never switches to ATTI mode anymore.

NP2019 Posted at 5-13 06:13
I would check on the micro SD card first.

When I first flew my Spark, I always experienced switches to ATTI mode. And then I noticed that the issue's gone when I only flew around and didn't record anything. After I replace the micro SD card with the fastest I could find (Sandisk Extreme Pro), my Spark never switches to ATTI mode anymore.

Well, now THAT's an interesting observation.

S-e-ven Posted at 4-23 18:03
You can't switch a bird, that is in real Atti, back to P-GPS. You just can switch a P-GPS bird to a mode, where it works like in atti.
It still is in GPS mode, has compass, no interferences, but is ignoring this, when the switch is used.
Aside, if using the atti switch to fly around like in sport (speed things up), you risk loosing the bird, in case you lose the RC connection. I do not expect, it switches the forced atti back, then.

I haven't tried it yet, but it would be worth the FARM mode. ATTI mode with RTH.
I think this is flying at ATTI and when the connection is interrupted, or you press the "come home" button then you return to GPS mode and return home. It's just my assumption.

I've read many of these cases last year. I think the change of direction at higher speed increases the chances of the compass getting confused. Spark has a bad ability to handle the problem. The strong wind and the sudden change of direction, which is the source of troubles, I think. The reason is that some people have never had such a problem that they control it differently. I've never had a problem because I use it for video recording and I usually fly with him slowly. I avoid sharp turns, windy weather, because this is not good for video recording. The Spark flyaway for those who think of a sports plane and play at high speed, playing with it. (my own opinion based on the reading experience)

Jakab Gipsz Posted at 5-13 12:27
I haven't tried it yet, but it would be worth the FARM mode. ATTI mode with RTH.
I think this is flying at ATTI and when the connection is interrupted, or you press the "come home" button then you return to GPS mode and return home. It's just my assumption.[view_image]

Perhaps you could "poke"  wolferl about the "farm mode"?
He may have tried that already?

https://forum.dji.com/thread-174400-1-1.html

S-e-ven Posted at 5-14 05:28
Perhaps you could "poke"  wolferl about the "farm mode"?
He may have tried that already?

The video link was not the correct link.
The setting of the sensitivity of the remote control was. (I've realized it for a long time and I use it)
But I found another Wolfgang Mahringer video that you wanted to show.
This does not show the FARM mode (9) but the ATTI mode (3).

Jakab Gipsz Posted at 5-13 12:27
I haven't tried it yet, but it would be worth the FARM mode. ATTI mode with RTH.
I think this is flying at ATTI and when the connection is interrupted, or you press the "come home" button then you return to GPS mode and return home. It's just my assumption.[view_image]

Hi Jakab,

I of course did try that "farm" mode while doing the ATTI mode video. Switching to farm mode (that is possible, it shows up as FARM on the app) it doesn't do anything on a Spark. It still flies like being in P-GPS mode.

Cheers,
Wolferl

Jakab Gipsz Posted at 5-13 12:37
I've read many of these cases last year. I think the change of direction at higher speed increases the chances of the compass getting confused. Spark has a bad ability to handle the problem. The strong wind and the sudden change of direction, which is the source of troubles, I think. The reason is that some people have never had such a problem that they control it differently. I've never had a problem because I use it for video recording and I usually fly with him slowly. I avoid sharp turns, windy weather, because this is not good for video recording. The Spark flyaway for those who think of a sports plane and play at high speed, playing with it. (my own opinion based on the reading experience)

Hi Jakab,

Yes, my Spark also likes to drop into ATTI mode when I fly it in strong winds. Your "rapid movements" theory seems plausible to me. It also explains why it happens more frequently with people flying the Spark in sport mode.
Personally I think it is a software problem. I do fly a NAZA-M V2 and a drone with an A2 flight controller, and both of them have that same ATTI mode drop problem, and I heard the same from other pilots flying DJI drones. I only fly these two birds very seldom now, because I feel a bit unsafe flying them. So I think of a "systematic" problem here.
DJI has hundreds of engineers, why don't they investigate into that and fix the problem, or at least tell us pilots what we can do to avoid problems.

Cheers,
Wolferl

Wolferl Posted at 5-14 07:20
Hi Jakab,

Yes, my Spark also likes to drop into ATTI mode when I fly it in strong winds. Your "rapid movements" theory seems plausible to me. It also explains why it happens more frequently with people flying the Spark in sport mode.

I think they know about the error, but they can't fix it or they don't want to fix it. I think he's a little bit afraid of having created Spark because many of them have not been more expensive and this is not good for DJI.
That's why they don't mind if they fall like flies.

Here are some simple things you can try. Never turn on the spark holding it, sit it on the ground. Never wear a ring, watch, coins in pocket, car keys  ect. Never take off from anything metal or near anything metal including but not limited to concrete ( could have rebar) when you turn on the spark don’t have the controller near it. Fly in airplane mode if using a OTG cable. Don’t have your phone in your pocket or anything on that is Bluetooth. What I’ve mentioned has help me.

I did have a similar issue while i was in an open Rubber estate area. I couldnt explain what was going on. Later the same day after reaching my hometown , everything went well. There are some occasions the SPARK can be a nut at times with no explanation for its behaviour !! I believe this is because of some compass imu combined error during initial start. Even imu or compass calibration at that time didn't  fix the issue. !!

Sayhelloforme Posted at 5-14 17:47
Here are some simple things you can try. Never turn on the spark holding it, sit it on the ground. Never wear a ring, watch, coins in pocket, car keys  ect. Never take off from anything metal or near anything metal including but not limited to concrete ( could have rebar) when you turn on the spark don’t have the controller near it. Fly in airplane mode if using a OTG cable. Don’t have your phone in your pocket or anything on that is Bluetooth. What I’ve mentioned has help me.

agreed and also to check UAV Forcast for the KP. Even on really sunny days the KP can be high. 1-3 is best 4 is good, 5 is not good to fly but you can do with caution. Just be prepared for drift BUT 6 or above is a no fly day from what i read and watched a few informative Youtube videos.

Jakab Gipsz Posted at 5-13 12:37
I've read many of these cases last year. I think the change of direction at higher speed increases the chances of the compass getting confused. Spark has a bad ability to handle the problem. The strong wind and the sudden change of direction, which is the source of troubles, I think. The reason is that some people have never had such a problem that they control it differently. I've never had a problem because I use it for video recording and I usually fly with him slowly. I avoid sharp turns, windy weather, because this is not good for video recording. The Spark flyaway for those who think of a sports plane and play at high speed, playing with it. (my own opinion based on the reading experience)

I agree with you somewhat, but that is not always the case.  I had two flights which were very manic.  1st time it happened, I suddenly lost connection, relatively close to RC no obstructions between.  The AC initiated a RTH due to signal loss.  Once regaining the signal, I tried three times to cancel RTH, and it ignored my command (it actually showed "failed to RTH" in the logs).  Throughout the 8 seconds after the last RTH attempt, there were speed errors, followed by yaw errors, then switching to ATTI.  The quadcopter was traveling straight when this happened.  I was finally able to manually control it again and eventually the error went away and it switched back to P-GPS.  
2nd time it happened, I was transitioning from hovering to forward flight up to 14mph, and 6 seconds later the yaw error/ATTI occurred again.
I then took it back up a 3rd time and tried my hardest to replicate the issue by switching to sport mode, and banging the sticks around as hard as possible trying to bring on the error, but was unsuccessful.  
Keep in mind there was very little wind that day.

SlowSL Posted at 5-20 05:55
I agree with you somewhat, but that is not always the case.  I had two flights which were very manic.  1st time it happened, I suddenly lost connection, relatively close to RC no obstructions between.  The AC initiated a RTH due to signal loss.  Once regaining the signal, I tried three times to cancel RTH, and it ignored my command (it actually showed "failed to RTH" in the logs).  Throughout the 8 seconds after the last RTH attempt, there were speed errors, followed by yaw errors, then switching to ATTI.  The quadcopter was traveling straight when this happened.  I was finally able to manually control it again and eventually the error went away and it switched back to P-GPS.  
2nd time it happened, I was transitioning from hovering to forward flight up to 14mph, and 6 seconds later the yaw error/ATTI occurred again.
I then took it back up a 3rd time and tried my hardest to replicate the issue by switching to sport mode, and banging the sticks around as hard as possible trying to bring on the error, but was unsuccessful.  

This is interesting!
So sure there are 1 pc compass built in, but it is not too reliable. The common mistake is that it is not the main problem that it enters into ATTI mode, but because of some software (firmware) error, the connection and control are interrupted, which in this situation (ATTI) may lead to the loss of the drone.
The DJI does not comment on how the Spark firmware handles the compass bug and what causes these errors.
It's sad and that's why I always fly with a spasmodic stomach, though it could be a fun recreation if everything worked as the ad says ...

Jakab Gipsz Posted at 5-20 10:35
This is interesting!
So sure there are 1 pc compass built in, but it is not too reliable. The common mistake is that it is not the main problem that it enters into ATTI mode, but because of some software (firmware) error, the connection and control are interrupted, which in this situation (ATTI) may lead to the loss of the drone.
The DJI does not comment on how the Spark firmware handles the compass bug and what causes these errors.

I've also watched flights on Youtube and read many threads on this problem.  It seems to be a fairly broad range of symptoms following the error, but almost all of them are due to the compass yaw error.  Most of them appear to be slowly traveling in one direction, not in sport mode when the error occurs.  Most of them seem to be far out into the country, no houses, buildings, etc.  There is always at least a jerk/twitch when it happens, but sometimes it yaws wildly, and rolls to one side or another.  It appears that the pilot tries to correct, but has a hard time keeping it leveled.  I can't tell for sure what the pilot is doing because there were no stick overlays on the videos, but from the threads I've read, they say it was really hard to control and some instances laggy connection and/or extreme control lag.  

As far as flyaways go, I'm still trying to figure out the root cause.  I know a lot of instances are proven to be pilot error, not having the skills to manually fly in ATTI mode and bring it back to their location, but some of them appear to have a mind of their own.  They claim that they just head off into the sunset, never to be seen again, but I wonder how many of them are simply drift due to high winds while stuck in ATTI, or maybe accidentally controlling the spark in the wrong direction and out of RC contact.  

Since the last time I commented here I had this issue occuring 3 times. Last one was the worst because it ended up in signal loss and RTH that I have never experienced before. No wind, far away from everything, up in the mountains. I am very fed up with Spark now, and I am also very angry of DJI that they don't have the stomach to comment on this.

I just looked at my log, I can confirm that when the  yaw error occurred, despite me manually pointing toward my home location and pitching forward, the quadcopter did keep rolling toward one direction (about 100 deg. off from the home point) at a speed ranging from 5-10mph for about 11 seconds.   At that point, it lost connection and drifted downwind for another 11 seconds before regaining connection.  
***edit***
Looked at my 2nd log.  When the yaw error occured, I centered the stick, and it kept traveling at the same heading, about 10-18mph for 8 seconds before stopping.

This is a very strange issue, I will likely never purchase another DJI product due to this issue and DJI not even acknowledging it.  This will be the death of my spark for sure, as the only safe place to fly this is over a flat wide open field... I'd have more fun staring at the wall for 16 minutes.  Just a matter of time before it happens to me over a forest or water, just one yaw error/disconnect and it will be toast.

Dear DJI admins!

Please ask the engineers why there are many compass bugs, speed errors, and ATTI mode changes?
And more importantly, why is the connection interrupted, why it cannot be controlled.
Many SPARK owners want to get official answers!

Have any of you tried NP's suggestion of getting a fast SD card?

I'm curious about this as some NEVER have the problem and some ALWAYS have the problem. As an embedded software guy, I'm wondering if it's taking so long to record the video to the card that it's not getting its other work done.

Tentoes Posted at 5-23 06:01
Have any of you tried NP's suggestion of getting a fast SD card?

I'm curious about this as some NEVER have the problem and some ALWAYS have the problem. As an embedded software guy, I'm wondering if it's taking so long to record the video to the card that it's not getting its other work done.

Hmm, I'm not sure that this would be my problem, however I WAS recording when it happened.  I've had 2 of 60 flights with the compass error.  I'm currently using a genuine Samsung Evo select U3 card.