Hi everybody,

I have learned that the spark sometimes appear to suffer from compass and loss of gps problems and today I almost lost my spark.

Approx. 300m away I turned the drone around to take it back home, it became somewhat unresponsive and I got the messages compass error and the exit p-gps mode (switching to atti). I was able to remain in control of the spark and was to navigate back using the spark camera. Less than 100m from home gps came back and everything appeared ok. Some seconds later I got a gps or compass malfunction message and spark was back in atti mode (appeared to make a quick drift to begin with, but I´m not 100 % sure I wasn´t to blame). Once again I managed to control the spark and land safely.                                 
When losing gps and compass the 1st time, the drone was at a height of approx. 70-80m and had contact with 21 satellites (no metal/constructions close, the drone was over forest land). I've gotten some errors over the time owning the spark, but nothing like this and so far away. The question for me now is to open a service case or not? I bought the drone last black friday and hence warranty is expiring soon (I do also have dji care). The drone, rc, battery and ipad have the latest firmwares. Do you have any experience on how dji have done in similar cases?

/Henrik

If compass goes, you can have 20+ sats, it is going to atti.
No compass, the bird ignores GPS
Upload your flight log to phantomhelp and post the link
That way people can tell you more.

Have you been flying in sport, when you lost the gps mode?

And check your imu and compass values in the app

I concur with Seven.. this is USUALLY not a compass issue... and its not limited to just he spark models.. my DUAL compass Mavic Pro has done the atti dance as well as 2 different sparks I own (returned one as it was real bad)

looking not just at the flight logs but logs of the flight controller itself it looks more like this issue stems from how DJI deal with a separation of values in yaw (FC heading) and magyaw (compass heading) .. if (when) these values become divergent it appears that the programmed solution is to reboot and likely in doing so, reset the kalman and complimentary filters that keep track of drift/dampening offset tables .. when this navigational reboot is going on.. the atti mode is just a wing leveler and even if GPS is available it makes not attempt to use that information.   

solid state compass modules are very sensitive to magnetic fields and interference and can/will be effected for the duration of a flight if the aircraft is launched from or near sources of interference..  the algorithm that shows this interference may or may not keep you to of trouble in this regard so ensure you launch from a good area and I always look at the radar view immediately after takeoff to ensure where the aircraft thinks it pointing actually matches with reality .. this is a good first indication that something isnt right.. another is that if immediately after takeoff while rising to precision takeoff height and/or going into a maintenance hover briefly before jetting off.. if the drone does a prolonged very slow rotation without input from the pilot .. this is the FC making good a difference in yaw magywaw trying to get them into agreement... you should be wary for the rest of the flight or land and investigate why immediately after launch its going that..

both the spark and to a lesser extent the mavic show a heading difference after more than 5 minutes of maneuvering flight.. less if in sport mode.  This started after updates from mid 2017 firmware to late 2017 early 2018 firmware and its been postulated that either additional features and modules added have caused a slow down in the processors ability to keep up with heading filtering/calculations or DJI purposely made firmware changes that have changed this algorithm in an attempt to smooth out previous reports of yaw and heading issues but in reality have made it worse..

any flight made with these products to the edge of VLOS  or in high winds is a elevated risk flight that could terminate in a lost drone.. plan accordingly

in the case of my atti episodes.. they have happened close enough to the RC that I did not suffer loss of RC connection or video telemetry loss.. although inputs were more sluggish.. but others have reported especially if this happens at further distances that immediate loss of all RC and video accompanied the atti, some had it come back, for some that was the last they saw of their aircraft..

some at DJI have acknowledged the issue in posts that go back to Sept, online and telephone support staff that I spoke too played dumb about the issue and spewed the typical re-flash firmware and do calibrations lines which really don't help in most cases unless the DJI app is asking for it.. its likely not out of calibration and  I attempted several on one spark and it would always RTH at heading deviations that approached 45 degrees off heading.. completely unacceptable.. it got returned to where i bought if from rather than go through DJI... if your drone is beyond sending it back.. and its under warrantee send it back to DJI and let them start dealing with this issues.. when it starts costing the company enough money they will perhaps be motivated to find the issue and patch it.

Thank you DeuceDriv3r for the very detailed explanation.

I don't really understand why DeuceDriv3rs response got downvotes. It's all correct what he says.
Since that flyaway problem started with the very first flight controls made by DJI (remember the DJI Naza M and Wookong M?), I do think its a business thing and we will never see a firmware correction for that.
That's a shame really.

Cheers,
Wolferl

I agree with you Wolferl, I think DeuceDriv3rs gave the best explanation so far of sensors problems that I have read on this forum or elsewhere. Thank you DeuceDriv3rs

DeuceDriv3r Posted at 11-17 06:08
I concur with Seven.. this is USUALLY not a compass issue... and its not limited to just he spark models.. my DUAL compass Mavic Pro has done the atti dance as well as 2 different sparks I own (returned one as it was real bad)

looking not just at the flight logs but logs of the flight controller itself it looks more like this issue stems from how DJI deal with a separation of values in yaw (FC heading) and magyaw (compass heading) .. if (when) these values become divergent it appears that the programmed solution is to reboot and likely in doing so, reset the kalman and complimentary filters that keep track of drift/dampening offset tables .. when this navigational reboot is going on.. the atti mode is just a wing leveler and even if GPS is available it makes not attempt to use that information.   

Thanks for a very detailed reply!

Some more info regarding the problematic flight. Yes, I had just switched to sports mode (does that increase the chances of getting compass error etc?) Another complicating factor was that I was at the border of the set max flight distance, possibly adding to any conflicts in data processing.

So given the info I've gotten it appears that  Sparks may have intermittent issues, some cases worse than others. It is a tough call when it comes to starting a service ticket or not since a new or repaired spark will most likely not be 100% problem free.

/Henrik

Same thing happened to me this weekend too, it is not a pleasant experience!

DeuceDriv3r Posted at 11-17 06:08
I concur with Seven.. this is USUALLY not a compass issue... and its not limited to just he spark models.. my DUAL compass Mavic Pro has done the atti dance as well as 2 different sparks I own (returned one as it was real bad)

looking not just at the flight logs but logs of the flight controller itself it looks more like this issue stems from how DJI deal with a separation of values in yaw (FC heading) and magyaw (compass heading) .. if (when) these values become divergent it appears that the programmed solution is to reboot and likely in doing so, reset the kalman and complimentary filters that keep track of drift/dampening offset tables .. when this navigational reboot is going on.. the atti mode is just a wing leveler and even if GPS is available it makes not attempt to use that information.   

I am in agreement with DeuceDriv3r’s very detailed analysis above. It has long been a hypothesis of mine that due to the relative short wheelbase of the Spark and it’s ability to make sudden and extreme course changes, especially in Sport Mode, that the FC and the Compass module may end up out of agreement with each other. The Spark will then revert to ATTI mode until such a time as the readings from the two sensors fall within a tolerance as determined by DJI.

My only experience of this is putting a GPS sensor on a race quad. In many cases the resultant data quite often plots the quad as though it was rolling to the side, with no change in heading, when in actuality it was a sharp yaw and roll. This is to be expected though as the GPS module ( FrSky S-Port GPS module) didn’t have a magnetometer and requires X# of data points before a change in heading is realised.

These are just my very basic observations and as such are open to debate, revision, or even flat out rejection.

Regards

Nidge

the question is WHY this yaw vs magyaw issue is more problematic now with the latest firmware than with the mid / early 2017 firmware

it would be interesting to see if there is a way processor load averages could be viewed pre/post firmware updates to see if there really is more processor involvement in the new vs old code.  the values may be accessible from the .dat files on the aircraft if csvview so the next time I analyze a file I will poke around and see if its a value that is recorded.


I am still betting on a change that was made to the flight controller modules of the firmware .. this really smacks of a filtering issue in the kalman or complimentary filters.. either the smoothing algo is too dampened/slow or the sensor fusion algo is either not updating samples quick enough or not weighing the compass enough to keep it in line/update  inertial/gps sources of heading...

another postulation is that the inefficiency of the flight controllers algorithms in coincidental with DJIs attempts in firmware hardening to limit access to 'tuning' parameters that people were over riding for no fly, altitude, etc and that the further encryption of the system might be to blame for the slowdown or bug.. its plausible ..

again DJI tend to have lots of common code base and since the new firmwares with issues started about the same time the new products were released... its coincidental at best but hunch lies there...

Am I the only one to notice?
The Admins NEVER report back the causes for problems that are reported common to many users. NEVER any feedback from DJI corporate.
Many times they carry on as if it is the first time a problem has ever been reported.

Hi Chuck101,
Maybe some work is been done, but we don't know what goes on behind the scene.

I think to put the record straight, Compass error speed error mag Yaw Yaw error, when Spark was first released in 2016 June , all of these problems were 10 fold when Spark was first released, and this can be explained in some way by a whole new set of drone flyers that came to the market , because Spark was targeted at new flyers, this resulted in many more problems and crashes than we now see in 2018, as someone who owned spark since its release and has been on this forum longer than Spark has, I can tell you that in the first 6 months of release all of these problems were much greater by a long way than they are today.
FW/SW are not the problem and dji has not said it was the problem as anyone who comes to this forum knows.
This is a repeat of the release of Mavic in 2015 when it was released compass error speed error Yaw mag Yaw etc and again 10 fold of what it is today and yes FW/SW was also bandied about but proven to be rubbish, Mavic and spark have never receive FW/SW update to either fix or sort this problem simply because it’s not a FW/SW problem.
Why do we see much less problems and crashes on both Spark and Mavic Forums, simply initially huge boost in new users and sales and now much more experienced users of both craft, and now we see much less problems with both Spark and Mavic,
If people were to go back to 2016 and 2015 they will clearly see what I’m saying, if they think that some magic FW/SW is coming there dreaming because it’s simply nothing to do with FW/SW.
If users take the time to do preflight checks andfly in good envoirments , we will see a lot less of these problems. Just my twopence worth.

Chuck101 Posted at 11-19 06:02
Am I the only one to notice?
The Admins NEVER report back the causes for problems that are reported common to many users. NEVER any feedback from DJI corporate.
Many times they carry on as if it is the first time a problem has ever been reported.

The are bots or what I call 'flesh bots' .. NPC characters reading off a script of approved answers.. you would get better information talking to the tree in your yard..

DeuceDriv3r Posted at 11-19 08:17
The are bots or what I call 'flesh bots' .. NPC characters reading off a script of approved answers.. you would get better information talking to the tree in your yard..

Very well put.

DeuceDriv3r Posted at 11-19 08:17
The are bots or what I call 'flesh bots' .. NPC characters reading off a script of approved answers.. you would get better information talking to the tree in your yard..

There are also 'flesh bot' deflectors / apologists.

HedgeTrimmer Posted at 11-19 13:58
There are also 'flesh bot' deflectors / apologists.

those are undercover operative NPC bots... hehe

Chuck101 Posted at 11-19 06:02
Am I the only one to notice?
The Admins NEVER report back the causes for problems that are reported common to many users. NEVER any feedback from DJI corporate.
Many times they carry on as if it is the first time a problem has ever been reported.

That is the exact reason for why I wrote the original post. It appear DJI do not want to open this "can of worms".

Still, I will keep Sparky and will from now on be careful not to enter sports mode when 300m away.