First, I apologize to everyone because I am totally aware there have been several posts on this topic before. I'd like to add a new thought to see what the group thinks. First, I'm aware of all the feedback that goes with flyaways and have reviewed several posts on how to try an minimize them. I'm not a novice pilot, I've been able to easily bring the Spark back during compass or satellite issues and land it safely. I have a pre-flight routine I follow before taking off involving satellites, IMU, compass, location, etc and I'd rather discuss what I've been noticing since purchasing this Spark on Jan 14, 2018:

• When the Spark is in the air for several minutes it is VERY susceptible to no or weak GPS signal strength. This happens mid-flight after sensors are all in the green before launch.
• What I noticed is if I am recording video, put the Spark in sport mode, and go full-throttle I sometimes get horizontal lines disrupting the video. I reviewed another post where a "pilot" was noticing similar issues with a Mavic. One user suggested this is prop shadows, and I disagree and I'll explain why.
• I noticed the Spark can have several compass errors mid-flight. I'm aware of calibrating the compass and all the nuances that go with it. This, again, is after it checks out great during the pre-flight checks.
• Anomalies 1 - 3 listed above nearly always only happen in sports mode.
  

Needless to say, as I mentioned, I'm not a novice pilot. During the last flight, the Spark reported a compass error followed immediately by a magnetic interference in which the Spark took off in a random direction not responding to the remove control whatsoever. Inevitably it "landed" on the side of the mountain. I wasn't able to recover the Spark until the next day, we had to hike to get to it. Oddly I cannot seem to export / download the "Blackbox" or the upload a "Flight Log" using the DJI Assistant the app "activity icon" just simply spins showing nothing to upload or export and I'd love to see the data.


Here is what I do know. There seems to be a load issue with the Spark when simultaneously recording video while giving it full-throttle in sport mode. It doesn't happen all the time, and the days I've flown it hasn't been windy when I'm trying to record. The day of the magnetic interference (MI) crash there was less than 5 Kmph windspeed, it was a beautiful day at about 70 degrees. I'm working with DJI Support to replace the Spark. Now that this has happened I have no trust in the Spark, I can't collect footage in locations I once used. I'm going to spend a lot of time testing the Spark under stress. It appears from the little bit of information that I've gathered that the Spark may be causing its own interference either as a resonant frequency issue or load issues under stress. Has there been any discussion on load and resonant frequency interference?

You could be onto something, it does seem like many of the problem children acted up during sports mode.

What was the K Index when that happened? I always use UAV Forecast before I fly and it takes all of that into account. Also, when flying in sport mode, try opening the battery info to see if there are any voltage drops while maneuvering.

I am trying to follow this problem.
Several reports of problems after high speed sport mode seem suggest this.
I tentatively suggest this is not magnetic interference, though the error flagged is yaw/compass errors.
Possibly the speed followed by a turn throws the gps/compass data, perhaps because of inertia.
This is nevertheless beyond my resources to prove.
Some seem to relock gps, some don't.
Voltage drops are another possibility.

PS I've not reproduced this myself. Maybe the gps is too slow to keep up.Flight log should record voltage drops.
Today's other example of this saw no voltage drop in the log from phantom help.

I have the same problem! I have done everything well: callibrated the compass and the IMU properly (I had some trouble calibrating the compass) and still the Spark flies away in an uncontrolled way. I have crushed it several times. It flies away at high speed and cant control it. A few times I managed to bring it back but the others it just crushed. I am really worry about this. It is not safe to fly this drone.

AdamTheGreat Posted at 2018-1-29 12:24
What was the K Index when that happened? I always use UAV Forecast before I fly and it takes all of that into account. Also, when flying in sport mode, try opening the battery info to see if there are any voltage drops while maneuvering.

Don't laugh, but I'm assuming you mean the Planetary K-Index. To answer your question according to NASA at 2:21 PM PST (UTC -8 or 22:21) the K-Index is between 1 and 2. If you didn't mean Planetary K-Index and there is some other reading I should have taken from the Spark that I'm not aware of please let me know.

djiuser_XPB97kItaqG4 Posted at 2018-1-29 14:34
Don't laugh, but I'm assuming you mean the Planetary K-Index. To answer your question according to NASA at 2:21 PM PST (UTC -8 or 22:21) the K-Index is between 1 and 2. If you didn't mean Planetary K-Index and there is some other reading I should have taken from the Spark that I'm not aware of please let me know.

Yes he is talking about the Planetary K index which can be indicated via an app and the website uavforecast.com

Bright Spark Posted at 2018-1-29 12:40
PS I've not reproduced this myself. Maybe the gps is too slow to keep up.Flight log should record voltage drops.
Today's other example of this saw no voltage drop in the log from phantom help.

I'm going to try once I have the Spark repaired or replaced. This could be a dangerous or troublesome situation, from all accounts this last area should have been about as good as it gets. The really irritating issues are I can't seem to get the black box or flight log data from the Spark. I'd love to take a look at the data.

fans8f69c869 Posted at 2018-1-29 14:20
I have the same problem! I have done everything well: callibrated the compass and the IMU properly (I had some trouble calibrating the compass) and still the Spark flies away in an uncontrolled way. I have crushed it several times. It flies away at high speed and cant control it. A few times I managed to bring it back but the others it just crushed. I am really worry about this. It is not safe to fly this drone.

Sir, sorry to hear that. Did you contact our support and start a case: http://www.dji.com/support . Please provide the flight records to them and locate the problem. Thanks for your support.

djiuser_XPB97kItaqG4 Posted at 2018-1-29 14:34
Don't laugh, but I'm assuming you mean the Planetary K-Index. To answer your question according to NASA at 2:21 PM PST (UTC -8 or 22:21) the K-Index is between 1 and 2. If you didn't mean Planetary K-Index and there is some other reading I should have taken from the Spark that I'm not aware of please let me know.

Sir, please provide me the case number and I would check the status for you. Sorry for your unexpected experience, we would like to help you.

I think you may be on to something.

Just got my spark a week ago, and yesterday and today I started to push sport mode quite a bit. I also started filming at the same time.

Out of the 4 flights, according to phantom pilots log viewer, each flight had errors.

#1 - Compass Error
#2 - Compass Error + Yaw Error
#3 - Speed Error
#4 (Today) - GPS Position NoMatch

So it looks like under sport mode I can pretty much force my Spark to do SOME sort of error on every flight. I just pulled out my SD Card and will try some aggressive flying without the SD Card and report.

I'm going to include a screenshot of todays flight with the GPS error. One thing that concerns me is notice the voltage on the batteries GO UP? How can this be normal...

Here is another one.

Looks like voltage drop This was in the same flight BTW... (#4 - Todays flight).

I'll add its fairly cold out today... But within DJIs stated operating temperatures. Temp right now is about 0-1C.

I think the OP may have hit on something with regard to the increased magnetic interference issue when in sport mode.
I use APM and Pixhawk controllers in my larger robotic minions. In the setup of these flight controllers there is a routine to calibrate against magnetic interference from the brushless motors for different throttle values. Essentially as the throttle is increased the flight controller measures the strength of the magnetic field created by the brushless motors and applies this to the amount of offset the compass experiences and creates what I presume is some form of truth table to compensate for these offsets.

In practice I have witnessed a very marked difference and improvement between the performance of the flight controller prior to and after this calibration routine is performed.

Unfortunately I don’t have a machine as small as the Spark fitted with one of these flight controllers to see if the magnetic field calibration produces the same beneficial results.

Ok. As mentioned, 4 previous flights... All shown an error of some sort.

Just did a flight with no SD card... 100% throttle through about 70% of the flight. Sport mode.

Not a single error.

Only difference was the sd card was off and I did not venture that far off. Stayed within 400 meters.

Still have about 2 hours of day light, will try again and report.

Early to tell but so far looking like having the video recording feature on and running at high throttle in sport mode is causing issues for me.

Nidge Posted at 2018-2-4 05:07
I think the OP may have hit on something with regard to the increased magnetic interference issue when in sport mode.
I use APM and Pixhawk controllers in my larger robotic minions. In the setup of these flight controllers there is a routine to calibrate against magnetic interference from the brushless motors for different throttle values. Essentially as the throttle is increased the flight controller measures the strength of the magnetic field created by the brushless motors and applies this to the amount of offset the compass experiences and creates what I presume is some form of truth table to compensate for these offsets.

I agree. My second quad is a TBS Discovery frame running a Pixhawk.

Running compassmot is essential.

Anom3 Posted at 2018-2-4 03:45
I think you may be on to something.

Just got my spark a week ago, and yesterday and today I started to push sport mode quite a bit. I also started filming at the same time.

About the battery voltage: I think (and I'm pretty sure about that) the battery voltage varies quite a lot during a flight, depending on the load, and this is completely normal. If the board and the 4 motors demand a lot of power the voltage will go down considerably even when you're at a 100% SOC, but it will also go back up again when less power is drawn.

So, 4 flights with no sd card, all in sport mode... 0 compass errors.

I do still get the yaw error, but I figure that's just from the aggressive sport mode flying.

DJI Elektra Posted at 2018-2-4 02:44
Sir, sorry to hear that. Did you contact our support and start a case: http://www.dji.com/support . Please provide the flight records to them and locate the problem. Thanks for your support.

DJI Support is taking great care of me. The Spark is set to arrive at DJI tomorrow. Thanks for responding, but I'm good at the moment. The case number is CAS-1448586-H5G5R7.

Anom3 Posted at 2018-2-5 05:58
So, 4 flights with no sd card, all in sport mode... 0 compass errors.

I do still get the yaw error, but I figure that's just from the aggressive sport mode flying.

If you take the SDCard out does the camera still run? If the camera isn't running then there will be less load. The processor on the Spark wouldn't work as hard, and there would no power draw from the camera. It takes a fair amount of processing to record which, I beleive, is part of the reason why the Spark doe not have a 4k camera.

Anom3 Posted at 2018-2-4 03:52
Here is another one.

Looks like voltage drop  This was in the same flight BTW... (#4 - Todays flight).

We are missing some context here.
Would be interesting to see throttle levels at time when voltage drop occurs.
I suspect motors might be drawing a bit too much current sometimes, certainly over the C rating of the battery causing such a voltage drops. And, what's the C rating of DJI batteries? I can't seem to find that data anywhere.

ktsummey Posted at 2018-2-5 06:18
If you take the SDCard out does the camera still run? If the camera isn't running then there will be less load. The processor on the Spark wouldn't work as hard, and there would no power draw from the camera. It takes a fair amount of processing to record which, I beleive, is part of the reason why the Spark doe not have a 4k camera.

After the crash, before I was asked by DJI Support to send the Spark to the service center I took it back out to stress test it. I flew the Spark in sport mode, while recording, intentionally trying to generate the same errors. I recalibrated the IMU and compass along with my other flight checks prior to flying. I was not able to reproduce the errors. This was in a location I've been to multiple times and have seen weak GPS signal strength after take off. I'll do a lot more testing once the Spark is back. I have to build trust back up that it won't fly away.

Anom3 Posted at 2018-2-5 05:58
So, 4 flights with no sd card, all in sport mode... 0 compass errors.

I do still get the yaw error, but I figure that's just from the aggressive sport mode flying.

Thats interesting theory regarding sports mode+filming. Im having the compass erors almost every flight and yes, the camera rolling all the time. My thought was that is has to something with cold weather and inertia of braking and sharp turning in sports.

eYeSkYeYe Posted at 2018-2-5 06:51
We are missing some context here.
Would be interesting to see throttle levels at time when voltage drop occurs.
I suspect motors might be drawing a bit too much current sometimes, certainly over the C rating of the battery causing such a voltage drops. And, what's the C rating of DJI batteries? I can't seem to find that data anywhere.

The context was set in the initial post, but here is a summary (long) to make it all easier. This thread was created due to a fly-away in an area free of anything. It's a wide-open park, free of any man-made structures and power lines. Prior to each flight include a pre-flight process that includes a GPS check, sensor check (IMU/Compass), RTH height evaluated and set, and return to home set. If I cannot get the sensors where I'm comfortable I won't fly in that location. I've actually had to pack and leave a couple of times.
Prior to the incident, I've been monitoring (learning to trust) the Spark, but nothing official or scientific. However, using sport-mode to record a "fly-by" video at full-throttle I've noticed the following:

• The resulting video can have these slowly to quickly scrolling horizontal lines. I can see the difference between prop shadows and these lines but I haven't totally ruled out prop shadows, shutter speeds (very bright day), or a voltage load. I don't have any ND filters yet so the shutter speed could be off the chart and I haven't verified that yet.
• The Spark will "lose" GPS signal mid-flight more often in sport-mode, but it has happened in a non-sport mode without even being a full-throttle (1.5 - 2.0 m/s)
  

The incident was not the first time I read "Compass Error", but it hasn't happened often. This was the first time I read a "Magnetic Interference" error. I'm not a novice pilot and I can confirm that I had zero control over the Spark when it accelerated as if I hit full-throttle in sport-mode doomed to "land" into the side of a mountain.


Upon reviewing the information I have I purposed the following possibilities:

• There is a system load issue @ full-throttle in sport-mode while recording video
• The brushless DC motors are causing the magnetic interference with the compass at full-throttle.
  

Before I was asked by DJI Support to ship the Spark to a DJI Service Center I took it back to a park where I have experienced minimal issues flying and did a stress test and did not get the weak GPS, compass error, or magnetic interference error. Even at this park, I have received "weak GPS" alerts mid-flight. The only issue I was able to reproduce are the horizontal lines in sport-mode. I'm not referring to seeing the propellers going from a hover to top-speed (full-throttle) in the upper left and right corners of the video, I'm referring to vaguely visible horizontal lines. I'm not ruling out the fact I'm making the camera work to hard trying to manage the extremely bright conditions here in Nevada. I think once I have my 6-pack of ND filters that issue may go away.

That is kind of a detailed summary.

ktsummey Posted at 2018-2-5 11:02
The context was set in the initial post, but here is a summary (long) to make it all easier. This thread was created due to a fly-away in an area free of anything. It's a wide-open park, free of any man-made structures and power lines. Prior to each flight include a pre-flight process that includes a GPS check, sensor check (IMU/Compass), RTH height evaluated and set, and return to home set. If I cannot get the sensors where I'm comfortable I won't fly in that location. I've actually had to pack and leave a couple of times.
Prior to the incident, I've been monitoring (learning to trust) the Spark, but nothing official or scientific. However, using sport-mode to record a "fly-by" video at full-throttle I've noticed the following:

After such a long, insightful and very useful post, I really have a heart-breaking time telling you that I was referring to Anom3s log.
Nevertheless, thanks for detailed explanation. And your theory about Spark's motor jamming Spark's compass is very interesting one.

For 15 mins flight, the spark is drawing 4 - 5 times the c rating, which is about 6 amps or 65- 70 ish watts.
For lithium bats that is a very modest load. Even doubling it to 10 C is modest, so voltage drop is unlikely with good batts, but cold may push it over  the edge I suppose. If it is dropping that much, maybe RTH is nuisance triggered.

Bright Spark Posted at 2018-2-5 12:09
For 15 mins flight, the spark is drawing 4 - 5 times the c rating, which is about 6 amps or 65- 70 ish watts.
For lithium bats that is a very modest load. Even doubling it to 10 C is modest, so voltage drop is unlikely with good batts, but cold may push it over  the edge I suppose. If it is dropping that much, maybe RTH is nuisance triggered.

Problem is we don't know C rating of DJI batteries. I mean, it's not uncommon for acro birds to have 70+ C rated batteries but I am now curious why DJI "hides" that piece of information. And it's actually peak load that matters... as these yaw and other errors happen briefly most of the time unless they escalate horibbly sometimes, it we can trust such a reports in the forum.

In spite of pushing my spark as far as I dare, I've never had an error.
But the possibilities  are many. Location, crash damage/abuse  however superficially minor, faulty batt/component are a few.

Bright Spark Posted at 2018-2-5 12:32
In spite of pushing my spark as far as I dare, I've never had an error.
But the possibilities  are many. Location, crash damage/abuse  however superficially minor, faulty batt/component are a few.

It's more a lack of transparency from DJI that upsets me than those pesky yaw errors happening in rare cases in very specific scenarios.

Very interesting and enlightening thread .. One thing that i fear the most is a fly away.
But nobody has ever put a finger on a possible reason (excluding pilot error).

This theory falls in line with the many posts ive read regarding compass errors and loss of gps.

I hope DJI takes this theory in mind and does more research on it.
As for me, for now im staying away from sports mode and fly within line of sight.
Keep this thread moving with more experiences and research .

ktsummey Posted at 2018-2-5 06:09
DJI Support is taking great care of me. The Spark is set to arrive at DJI tomorrow. Thanks for responding, but I'm good at the moment. The case number is CAS-1448586-H5G5R7.

Thanks for your update. If you have doubt in your case, please contact us.

Case #CAS-1448586-H5G5R7

DJI Support has taken a major turn for the worse, I'm not quite sure how to share this without boring everyone to death. The crash happened on Jan 27th, and I contacted DJI Support on the 29th. They sent me a form to complete that included my all the info they requested through email. I uploaded my flight records on the 29th, February 1st, February 7th, 8th, and 9th. The DJI Service center received the Spark on Feb 6th, then by Feb 7th they claim they can find no evidence of a malfunction. After repeated phone calls I had one representative (Mark) refuse to allow me to speak with a supervisor. It's clear that support nor the engineer viewed the flight records. I have a DJI Affiliate account and my own business site the promotes DJI and I'm considering taking it all down. Support is horrible.  After 5 calls to support a supervisor finally called me back, again, very clear they had not looked at the flight records. So I made a video:

I now understand why there is such a low opinion of DJI Support.

A lot of this make sense to me. I have tried twice to fly in sport mode, full speed and take video. Both times I crashed into trees that were well far enough from my flight path. The air craft seemed to be almost pitched at an abnormal angle side ways and I was not in a radical or steep turn. Almost like it tripped over itself.

ktsummey Posted at 2018-2-10 13:16
Case #CAS-1448586-H5G5R7

DJI Support has taken a major turn for the worse, I'm not quite sure how to share this without boring everyone to death. The crash happened on Jan 27th, and I contacted DJI Support on the 29th. They sent me a form to complete that included my all the info they requested through email. I uploaded my flight records on the 29th, February 1st, February 7th, 8th, and 9th. The DJI Service center received the Spark on Feb 6th, then by Feb 7th they claim they can find no evidence of a malfunction. After repeated phone calls I had one representative (Mark) refuse to allow me to speak with a supervisor. It's clear that support nor the engineer viewed the flight records. I have a DJI Affiliate account and my own business site the promotes DJI and I'm considering taking it all down. Support is horrible.  After 5 calls to support a supervisor finally called me back, again, very clear they had not looked at the flight records. So I made a video: Magnetic Interference Malfunction

Stories like these scare the crap out of me for ever needing DJI support.  Clearly this is not the first time I've read a story like this online...with proof of a malfunction...and with DJI support being absent and customer service skills lacking.  They clearly have great engineering...but I think their customer service is going to ruin them in the end.

however to add my spark notes to this thread.  I have recently, and in very cold weather, flew the spark for at least 4 batteries in sport mode for at least 90% of the time, recording the whole time.  I was chasing Ice Boats on a frozen lake.  They were hard enough to keep up with at 30mph.  However other than range, I never had a problem with control.  I'm wondering what the speed was of the SD card in the above example while recording.  Maybe if it was too slow of a card it was causing a processsor to be sluggish? (THAT IS A COMPLETE GUESS)  No idea. Just throwing that out there.

It may be caused by CPU (and ESC controller) overheating during flights. The cooling ventilation in Spark may not sufficient because it does not have enough air intake vents. A Youtuber pointed out this fundamental issue in Spark. To avoid this problem, Mavic Air is improved to included more air intake vents to prevent CPU (and ESC) overheating.
CPU (and ESC) overheating may cause many weird problems during flight, e.g. lost connection, video lagging, uncontrollable flights, crash and flyaways. DJI, of course, will not accept this is design fault. :-). If DJI is a responsible company and treasures its goodwill, it should consider to recall ALL the Spark from the market.

If uncontrollable crash/flyaways continue to happen/report everyday in flying Spark, company's goodwill will not last long. Soon to become zero or even negative goodwill (in accounting terms).

I sent all this info to DJI Support. It's a complete safety issue at this point. It's sad DJI support is this bad.

Ricky Aerial Photography Posted at 2018-2-10 17:18
It may be caused by CPU (and ESC controller) overheating during flights. The cooling ventilation in Spark may not sufficient because it does not have enough air intake vents. A Youtuber pointed out this fundamental issue in Spark. To avoid this problem, Mavic Air is improved to included more air intake vents to prevent CPU (and ESC) overheating.
CPU (and ESC) overheating may cause many weird problems during flight, e.g. lost connection, video lagging, uncontrollable flights, crash and flyaways. DJI, of course, will not accept this is design fault. :-). If DJI is a responsible company and treasures its goodwill, it should consider to recall ALL the Spark from the market.

Agreed. I have noticed the Spsrk get very warm during use. Since mine is tied up with DJI Service I can't test it. I'll after that to the power load and resonance theories.

for me that's a pretty bad story - with all evidence there it should be no futher qqing from DJI. just replace a malfuncioning unit under warranty...