I've been flying my spark for about 4 months.  This was the first time I ventured beyond my own neighborhood to a nearby state park to fly.  I was flying with the RC connected to my iPad mini via cable.  

Just after 2 minutes into the flight I got an error message, it went by fast, but I think it said compass calibration error.  I lost contact and watched the drone fly off in the other direction beyond the trees and out of sight.  I pushed the throttle up so that in case it was reading any signal it would avoid the trees ahead of it.

I chased after it, and luckily when I got closer the RC apparently regained connection.  I could hear it but I couldn't see it.  Followed the sound of the drone and looked up and there it was hovering above the trees.

My theory is that the drone got caught in a wind gust and lost it's direction. Then it took off.  It flew in the opposite direction of the home point.
The ground windspeed had gusts up to 10mph, a speed which didn't seem prohibitive at the time, but I don't think I will fly in again.  

Any ideas, input on what happened and why?

<iframe width="560" height="315" src="https://www.youtube.com/embed/P3h8HiC0iJY" frameborder="0" allow="autoplay; encrypted-media" allowfullscreen></iframe>

Have you been/ did you just switch in sport, as it happened?
In the video it is either in sport or in Atti (which is kind of a sport mode, too)

Replay the flight record in the app
"it went by fast, but I think it said compass calibration error"

If it said "compass ...", it went to Atti mode, which is in the air above 5m a "drift with the wind" mode.
The meaning of that is that you would have to keep it with stick work at the position or best bringing it back to you, because the GPS will not be used then.

Rising the altitude, as you did, is a good move.
As the bird will mostly gain back the compass after some seconds of not maneuvering.
(Own experience, Atti is kind of like in sport, so maneuvers are harder, quicker, keeping the desorientated compass longer desorientated, methinks)
And then GPS gets back on and it will hang in the air, hovering till battery-RTH or critical-battery-landing kicks in.

Upload your flight log for this event at http://www.phantomhelp.com/LogViewer/Upload/ the paste the resulting URL back in a reply... we can then look at what happened.  All the instructions are there in that site on how to find your flight records on your mobile device.

Yet another compass / fly-away / un-controlled spark event...  what firmware were you on ??  .1000 ?

when a vehicle drops into ATTI ... it should not lean full tilt into some random direction and take off like a horse that just got whipped... it should hover, level, and sit there stupid only drifting in the direction of and at the speed of the prevailing wind.  

in ATTI it should have full stabilization but it will disregard navigation and not attempt to hold position as it has no sensor inputs that it currently trusts.... hence it drop from P into ATTI

S-e-ven Posted at 2018-10-13 20:16
Have you been/ did you just switch in sport, as it happened?
In the video it is either in sport or in Atti (which is kind of a sport mode, too)

Thank you for your response. No, I don't think it was in sport mode. But I did just update the firmware.

White Heron Posted at 2018-10-14 08:32
Thank you for your response. No, I don't think it was in sport mode. But I did just update the firmware.

so you were on firmware .1000

color me surprised ... sarc

how many more sparks have to go rogue and people / property hurt or damaged before DJI addresses and fixes this...

DeuceDriv3r Posted at 2018-10-14 07:03
when a vehicle drops into ATTI ... it should not lean full tilt into some random direction and take off like a horse that just got whipped... it should hover, level, and sit there stupid only drifting in the direction of and at the speed of the prevailing wind.  

in ATTI it should have full stabilization but it will disregard navigation and not attempt to hold position as it has no sensor inputs that it currently trusts.... hence it drop from P into ATTI

But most people are active on the sticks, when Atti happen.
You know what is happening, if you are flying in P or tripod, and switch to sport (Atti, in this case)?
It is going wild, compared to befores leaning and tilting.

In my experience, the best to handle a compass (atti) problem, is to not use the right stick for some seconds, just let it go. And the left just to get as high as possible.
Seems to me, is the quickest way to get it back under control (in GPS Mode)
Also I would suggest, if in Sport, switch back to P, minimum for a while.

S-e-ven Posted at 2018-10-14 08:57
But most people are active on the sticks, when Atti happen.
You know what is happening, if you are flying in P or tripod, and switch to sport (Atti, in this case)?
It is going wild, compared to befores leaning and tilting.

agreed and you are absolutely right if you are full on the sticks and drop into ATTI..

perhaps the code should be written that dropping or changing modes should ignore stick positions at the time of the mode change..

IE drop into ATTI... go zeros on stick input until the sticks are let go, return to zeros to reset then begin to fly...

but many report that the link is dropped and goes into ATTI at the same time but the recovered video from the CARD shows that the drone took off at full speed and never stopped... even after link to the remote/phone dropped...

the drone should default to a hover any time there is an interruption in control connection, change to or from a flight mode etc .. until the sticks are re-centered

DeuceDriv3r Posted at 2018-10-14 09:02
agreed and you are absolutely right if you are full on the sticks and drop into ATTI..

perhaps the code should be written that dropping or changing modes should ignore stick positions at the time of the mode change..

DJI?
Anyone reading?

S-e-ven Posted at 2018-10-14 09:23
DJI?
Anyone reading?

thanks for taking the time on the up vote... my first I think!!!

Well, the thing is the compass, IMU, YAW error was happening way before 0.1000 firmware and was not related to a firmware change.
All of the sudden, people with newer or newer exchanged Sparks started having this issue.
As if DJI changed the mainboard or component.

White Heron Posted at 2018-10-14 08:32
Thank you for your response. No, I don't think it was in sport mode. But I did just update the firmware.

In the video you will see sharp movements of the camera.
That is indeed Sport Mode.

When you are in normal GPS mode, the gimbal stabilizes the flight on two axis.
In Sport Mode, the gimbal is locked to 0°p and 0°r preventing damage to the gimbal.
This lock of the gimbal records motions very much like the ones in your video.

All we need is the flightlog....speculations is 1, data analyzing is another cookie   

@White Heron, if you like upload your flightlog > http://www.phantomhelp.com/LogViewer/Upload/

cheers
JJB

pmshop Posted at 2018-10-14 10:59
Well, the thing is the compass, IMU, YAW error was happening way before 0.1000 firmware and was not related to a firmware change.
All of the sudden, people with newer or newer exchanged Sparks started having this issue.
As if DJI changed the mainboard or component.

well with regards to my TWO sparks.. both were flying great for more than 30 flights on each.. kept getting harassed to update firmware so I updated both...

since I have 4 batts I updated both so that all birds, batts and remote would be on harmonious firmware

huge mistake

since updating - BOTH aircraft

make more prop noise in hover and especially in decent

much more unstable in decent.. the attitude indicator is all over the place

the aircraft does not track to the home point while pointing at the home point. the track is straight but flies back in 30 degrees off bore pointing way to the right of its flight trajectory.. no its not the wind..

the aircraft reports a different heading than its actually pointing in.. and when it takes off it precision lands angled off.. never did that before it would come back and rotate to the heading it took off from

the aircraft camera recording will yaw (appear to turn) while the aircraft is not turning . in the air in hover or on the ground

more disconnects..

less flight time

I have attempted several IMU and compass calibration all under ideal conditions to no avail...

I have not had it gone rogue yet but I ACTIVELY choose ideal conditions to fly in now.. I ensure 12+ sats, ensure good PDOP and HDOP, never above a Kp value of 3 and no solar flare activity above a 2... i.e. not even give it an excuse to break GPS lock... because if I could get it to actually report mag heading I know for sure it would be wrong..

I have thousands of flight hours behind real IMU/INS systems and hundreds of drone hours... and the latest updates have taken TWO superbly stable and well flying aircraft and degraded them severely..

DeuceDriv3r Posted at 2018-10-14 11:23
well with regards to my TWO sparks.. both were flying great for more than 30 flights on each.. kept getting harassed to update firmware so I updated both...

since I have 4 batts I updated both so that all birds, batts and remote would be on harmonious firmware

Completely opposite with my TWO sparks.
Both still behaving as they should (slight gimbal holding the horizon issues)
But both were last replaced in March this year.
I started seeing more and more of the described issues around May.
right in the middle of 0.0900 and 0.1000 releases.
If you are confident that 0.1000 firmware wrecked them...what is stopping you from rolling back to 0.0900?

I won't claim to be the genius with only 180 hours before I started keeping track in December of 2015.
All I can report is what I have expereanced.


By the way, Kp index is bogus.
I flew autonomous missions when the Kp Index was 7 back in a storm of 2016 with no issues - this was before I started hearing the arguement of the effect/ no effect.
Matter of fact, I flew tonight with a value of 5 without issue.

DeuceDriv3r Posted at 2018-10-14 11:23
well with regards to my TWO sparks.. both were flying great for more than 30 flights on each.. kept getting harassed to update firmware so I updated both...

since I have 4 batts I updated both so that all birds, batts and remote would be on harmonious firmware

So why dont you downgrade ones bird firmware?
The battery and the RC are working in the latest with the birds former.

S-e-ven Posted at 2018-10-14 18:01
So why dont you downgrade ones bird firmware?
The battery and the RC are working in the latest with the birds former.

Both of these sparks bought within the last month came with .701 firmware. Not .900

That DJI is still shipping the, with .701 should tell you something right there since that is a 2017 firmware

I hate downgrading firmware. Chances of bricking run higher and really dji should address the elephant in the room and fix these issues

http://www.phantomhelp.com/LogViewer/FRS0BK0MA2MQA9KDVZ94/

I have uploaded the flightlog here.

Thanks everyone for your input.Flight Log Data

Hi White Heron,

Try do understand your flightlog, its a flight of 7m 48 secs. Its last record is in Sport mode low altitude but no landing?  What happend.
Your video 1m 23 secs, can you tell wich part of the flight is shown?

Looks like all yaw stick inputs were followed by Spark.

No compass/imu errors in the log.

cheers
JJB

Food for thoughts....

So why cant experts and DJI representatives elaborate on solutions for these matters (mentioned above in posts) .

- why dji spark easily swayed by the wind
- is it impossible or difficult to find a solution for a drone to stabilize it self after such errors and not be carried a way with the wind, given the company resources and assets.
- is it a marketing strategy by not giving a perfect drone and segmenting solutions over future spark versions?
- are there other factors that can make this (compass errors, imu, gps...) and impossible job.

cheers

DeuceDriv3r Posted at 2018-10-14 18:14
Both of these sparks bought within the last month came with .701 firmware. Not .900

That DJI is still shipping the, with .701 should tell you something right there since that is a 2017 firmware

That may be a conicidence!

Just a month ago I saw a Combo Spark with .1000 out of the box, RC on latest version, too.
Build August 1, the serial check told us.

Aside, I have 2 birds, did upgrade first one to .1000 and just one RC (just in case), 3 month ago.
After having no problems, I upgraded the second RC , too, because it was just about to end the warranty time.  But haven't had trouble with that, too.
Just: Upgrade the bird with cable and computer!
And just the RC, after that, wifi connected to the RC
My second bird was in service, and came back upgraded to .1000. And it just works good, till now.

S-e-ven Posted at 2018-10-15 01:31
That may be a conicidence!

Just a month ago I saw a Combo Spark with .1000 out of the box, RC on latest version, too.

and you can confirm that after upgrade, running firmware .1000

that flying long distance back to home point in RTH that the home point location is centered in the video feed?

I upgraded one over the air with the iOS app and another using the latest version of DJI assistant.. both are having issues.

the aircraft were bought 2 weeks apart from 2 different suppliers

how do you do a DJI serial number check?

thanks

DeuceDriv3r Posted at 2018-10-15 04:18
and you can confirm that after upgrade, running firmware .1000

that flying long distance back to home point in RTH that the home point location is centered in the video feed?

serial ceck: http://tools.retroroms.info/

I can't be certain about the RTH thing, b/c I do not use that feature often, luckily ;-)
Not sure if I can do that in the next days, a lot of rain around, right now.
And not always time, when it is dry!

But just to confirm:
Force a reinstall on one bird, for .1000 with the assistant.
And downgrade, with assistant, the other to .900 (or .701, of course)
IMU calibration and compass calibration

And fly an RTH before and after.

S-e-ven Posted at 2018-10-15 05:03
serial ceck: http://tools.retroroms.info/

I can't be certain about the RTH thing, b/c I do not use that feature often, luckily ;-)

might give it a shot..  but I really hate messing with firmware more than I really have to.. just too many cases of bricking or corruption and not just in drones but everything..

when you don't use RTH, for example, than you really don't know what the drone is 'thinking'

its not like in a real aircraft where I can pull up INS maintenance pages and compare INU heading and compare it against the know magnetic heading or other nav sources

with the drone its failing the 'observed' tests on its ability to orient correctly.

i.e.  take off with it facing a specific direction, when its returned and put into auto land.. it should face that exact heading.. it no longer does... it looses that heading ref during the flight... sometimes right after takeoff

it could be changes to the complimentary or kalman filters, could be firmware in how the components are operated.. but I can't imagine that its hardware that just happened to change on 2 different aircraft first flight right after firmware changes.. the odds of that are in the millions

DeuceDriv3r Posted at 2018-10-14 11:23
well with regards to my TWO sparks.. both were flying great for more than 30 flights on each.. kept getting harassed to update firmware so I updated both...

since I have 4 batts I updated both so that all birds, batts and remote would be on harmonious firmware

You really watch the Kp index ? You are in Alaska  ? Is it really something to be carefull with ? I know solarflares can disturb radio communication. But living in Norway and being Alot outdoors using Garmin GPS, I never ever had problems with positioning. But maybe drones are more sensitive.

Anyhow, if the new FW gives yo so much trouble, ever considered downgrading ?

I think, or at least it looks like, that there are several different hardware versions around, all reacting different on FW updates.. Kinda strange thing to do.. Why not come with Spark 2  version.. Weird..

wimherman Posted at 2018-10-15 05:58
You really watch the Kp index ? You are in Alaska  ? Is it really something to be carefull with ? I know solarflares can disturb radio communication. But living in Norway and being Alot outdoors using Garmin GPS, I never ever had problems with positioning. But maybe drones are more sensitive.

Anyhow, if the new FW gives yo so much trouble, ever considered downgrading ?

it is difficult to know exactly, the effects on DJI products due to the closed nature of their flight control software and more importantly their filtering algorithms ..

in the military we have aggressive GPS satellite norming algorithms at are constantly computing satellite geometry, RAIM and DOP.. additionally tracking observed vs expected signal timing etc and will drop satellites from the 'solution' if they are not performing appropriately due to geomagnetic, reflective, multi-pathing, etc errors...

geomagnetic storms have been definitively shown to interfere with satellite communications and timing, and thus there are GPS implications as the GPS positioning system relies obviously on just that timing of signals from multiple satellites in known positions...

DJI drones use a combination of GPS, barro, MEMS IMU, compass, VPS sources though complimentary and kalman filters for navigation... so depending on this combining, weighting, filtering, etc of these sources vs the accuracy of the calculated navigational data really can't be analyzed since access to the raw values are not permitted in the firmware.

DeuceDriv3r Posted at 2018-10-15 05:24
might give it a shot..  but I really hate messing with firmware more than I really have to.. just too many cases of bricking or corruption and not just in drones but everything..

when you don't use RTH, for example, than you really don't know what the drone is 'thinking'

"when you don't use RTH, for example, than you really don't know what the drone is 'thinking"

I did use it, I checked it out, but I have no reason, to fly my birds back via RTH.
I (have) use(d) it, sometimes, to get the bird pointing back home.
What it then did.
But I mostly don't fly the Spark pointing direction to the homepoint for video reasons.
There is almost always something that is more interesting, methinks.

But I will look for it, next time I fly.

S-e-ven Posted at 2018-10-15 06:36
"when you don't use RTH, for example, than you really don't know what the drone is 'thinking"

I did use it, I checked it out, but I have no reason, to fly my birds back via RTH.

I don't USE it because I initiate it... it gets USED ON ITS OWN .. its COMMANDED by the DJI firmware when there is a interruption of 3 seconds with the RC

and when it auto commands a RTH it will plot a rhumb line direct to the home point recorded at takeoff then begin navigating that rhumb line.. all the while calculating observed navigation made good vs navigation requested.. and if it decides that there is degregated  performance it may drop into ATTI mode... then the fun really starts....

Yep, thats whats I mean:
I am lucky, most of the time, that I dont have RTHs initiated.
Or that they last long, if that happens.
And as soon I am back in control, I take control.
So I may just not have recognized the angle of the bird during RTH.

I will force it as soon as possible to check if your findings are reproducable with my birds.

if you are FCC .. get it out at least 300 meters... fly around a bit... maneuver ... sport mode a bit as well... then command a RTH and see

1: if the RTH point is centered in the video on the way back

2: the drone rotates and lands facing identically the same heading it took off from

DeuceDriv3r Posted at 2018-10-14 07:03
when a vehicle drops into ATTI ... it should not lean full tilt into some random direction and take off like a horse that just got whipped... it should hover, level, and sit there stupid only drifting in the direction of and at the speed of the prevailing wind.  

in ATTI it should have full stabilization but it will disregard navigation and not attempt to hold position as it has no sensor inputs that it currently trusts.... hence it drop from P into ATTI

Nothing hovers in a fluid, such as air, without GPS, pilot input, or some other stabilization method (optical?).  The Spark absolutely will not hover, in ATTI mode, but will always run off in some direction because what it thinks is level is never perfect enough to stay still in space.  

Just ask any full size helicopter pilot or model helicopter pilot about the machine automatically hovering itself without autopilot GPS lock.  They will find the question humorous after all the hours they spent just learning to hover the machine.

You just needed to input corrective action on the remote controller's sticks when the autopilot had an error (Compass/GPS loss).  Any time the aircraft isn't behaving, you have to correct with the sticks.  Don't trust that the autopilot will always correct the drone's flight path.

Gunship9 Posted at 2018-10-15 16:44
You just needed to input corrective action on the remote controller's sticks when the autopilot had an error (Compass/GPS loss).  Any time the aircraft isn't behaving, you have to correct with the sticks.  Don't trust that the autopilot will always correct the drone's flight path.

You are wrong and are completely missing the point to boot

And your constant comparisons to hell’s is a joke. So stop

I have flown both hell and non assist race quads. A spark in atti if you are off the sticks most certainly will be pretty dam level and hover just fine. I can do it in doors with vps turned off and its one of the most stable drones I have flown due to the canted motors.

But if you are full forward on the sticks in tripod going 2mph and it drops into atti you start going 30 as the quad fully leans into the commanded stick.

When dropping into atti it should hover hunting the sticks are re-centered so it does not take off like a scalded dog

In any non gps hove it will of course go with the wind but that is far different than 20 deg tilt full throttle

I had a simular incident last week, it was a touch windy but i lost all control for a bit until I told it to climb, one it did I had control back. I think the key is telling it to climb

if you had a strong gust of wind to take your spark for a ride  you would have had a message saying something like " danger strong wind "   ( or some message like that ) . i received a message like that  when i was flying in a strong wind last year .

DeuceDriv3r Posted at 2018-10-15 18:52
You are wrong and are completely missing the point to boot

And your constant comparisons to hell’s is a joke. So stop

You are wrong and your claims that an drone in ATTI will hover without a pilot controlling it is a joke..  So stop.

I like your claim that a Spark in ATTI "will be pretty darn level" and that is all that is needed to hover.  I disagree that that is all that is needed to hover without a pilot on the sticks.  A hover, still in space, requires constant corrections from the autopilot using GPS-P or vision systems (low level).  You claim that a hover can be achieved witout constant corrections?  

Perhaps we disagree about what a hover is.  Do you mean a hover is maintaining altitude while it flies off in a direction/into a wall?  A hover is a term related to helicopters so referring to helicopters is valid.

Interesting idea that the Spark should ignore the sticks when changing flight modes until the sticks are recentered.  Is this common operation with drones?  Do other drones do this?  Seems like it would befuddle the pilot.

Not a hover:

what I am saying is this.. you are in gps tripod mode full stick  going 2 miles an hour and have full stick input that is taking the drone toward a large immovable object .. lets say 15 feet away... all of a sudden.. it drops into atti... by the time you hear atti mode uttered by the controller .. the drone jumped into a full sprint 30mph collision with the wall before you even know why...

yes .. I believe that if you have stick inputs in and the drone changes flight mode.. the stick inputs should be ignored and the drone should hover.. station keep... not move if you are really dense... until you recenter the sticks .. recentering is like acknowledging the mode change and then the drone can respond to stick inputs again...


and that spark return gif.. if that is a RTH it looks nothing like how my spark autopilots back.. .tells me nothing about return heading vs heading to home point

My Spark was completely locked for months.  I couldn't believe how stable the drone was until some point this summer when I was flying it and got a compass error and the spark started to blow off course.  The drone went to ATTI mode and I had to fly it back to me manually.  By the time I got half way back it was back in GPS mode and stable again, but it would jump in and out of ATTI mode showing Compass and GPS errors even though it would often have a dozen or more satellites.  This started happening on every flight including locations where I know the drone was perfectly stable in the past.  This made "video shoots" very difficult which for me is the only point of the drone.  I was always able to control the Spark and bring it back safely every time but, I couldn't risk flying it too far away for fear I would lose control of the drone.  If it initiated a return to home AND lost GPS at the same time I thought it would just fly away somewhere.  I also was not able to fly in and out of a tricky spot (like launch from a boat, or sneak it up through a break in trees in the woods) because going into atti mode in a tight space could cause a crash.

After a dozen or so flights, I concluded there must be something wrong with my Spark so I sent it back to DJI using DJI Care.  I just got a report that GPS was not functioning correctly but it was being repaired for free under warranty.  (not sure if this is standard warranty or the 1 year DJI care I bought???  I assume the later.)

I did update to the latest firmware, but my issues started before the upgrade.  I hoped upgrading the firmware would fix my problem but it of course did not.   I'm not sure what caused my GPS errors but if you are getting the same errors, I would open a case with support.  If you have DJI Care, it will be covered.  If not, it looks like it might cost just under $80 to fix.