Hey folks,

This post is an extension to

https://forum.dji.com/home.php?mode=space&do=profile&mobile=2&uid=1415621&uuid=f451d2229e56b6?

where I posted a video of the situation. The drone basically crashes after takeoff when I push the left stick upwards.

A lot of speculations were written and DJI refused any responsibility, telling me it's the phones fault since it's not on the list.

Well, since last time I refreshed the firmware, did a IMU calibration and only used my girlfriends S5 since it's on the supported device list. Everything seemed to work fine until yesterday.

I did a palm launch like always and the drone hovered in front of me. I pushed the left stick upwards to let it gain some height. But instead of climbing the drone just went down into the beach and of course now it's not usable anymore since sand went into the motors.

Here is the flight log...

http://www.phantomhelp.com/logviewer/X871U2FMSK1RTR0DEHO5/

I'll open a repair case and keep you posted.

Best
Martin

When you pushed the left stick upwards to gain some height, your Spark did ascend (as expected). It did not just go down into the beach as you described above. After that point, it looks like the battery might have disconnected from your Spark mid-flight.

Wow, how far did it ascend? Do you see that VPS goes down while the IMU shows and upward direction? The motors were running when it crashed into the sand and the drone was still running when I picked it up. So it's probably not the battery.

If you watch the videos from the first thread you'll see what's going on.

What's this about a list with phones, can anybody link me to that list?

If you look in the old thread you'll find a graphic that DJI posted.

If you pull the DAT flight log from your Spark, you should find it contains all of the flight data (including the part where your Spark is on the ground). You can find instructions for retrieving that log file here.

I'll do that as soon as I have time for that. Thanks so much for that information.

msinger Posted at 2018-7-11 05:27
When you pushed the left stick upwards to gain some height, your Spark did ascend (as expected). It did not just go down into the beach as you described above. After that point, it looks like the battery might have disconnected from your Spark mid-flight.

I don't claim to fully understand the .csv file, but some things look odd.

The VPS altitude makes a sudden jump at 4.6s, probably when he removed his hand from under the drone.

He gave it throttle at 5.2s. The altitude column (from GPS??) shows it going up, the VPS altitude shows it going down, and the VelocityZ is negative. Something weird is happening. Could it be a poorly calibrated IMU? I'd like to know what the source is for the data in the various altitude columns.

if looking only to what has been described, can it be a different "control mode"? instead of Mode2 to be something else?

MarkMN Posted at 2018-7-11 07:59
I don't claim to fully understand the .csv file, but some things look odd.

The VPS altitude makes a sudden jump at 4.6s, probably when he removed his hand from under the drone.

Looks to me like the vpsHeight increased at 5.2 secs.

The altitude is derived from the barometer.

VelocityZ is another label for VelocityDown, and so it has the opposite sign from what you would expect

Good luck in getting DJI to do anything about it, they will blame the sand, see my thread that was moved out of discussion to service     

https://forum.dji.com/forum.php? ... 6orderby%3Ddateline

MartyTravels Posted at 2018-7-11 05:54
If you look in the old thread you'll find a graphic that DJI posted.

Can you insert the link to the old topic, not to your profil, please.

Tell you service guys from DJI, this :
"and DJI refused any responsibility, telling me it's the phones fault since it's not on the list."

Is rubbish!
First because there is no "not usable with (phone) list", the only requirement for android phones is android 4.4. And the list is just a recommended and testet list.
Second because I had this 2 weeks ago without a device connected to the RC.
That is aproximately a powerfailure somewhere in the IMU control for the motors.
I would bet, if you upload you log to phantom help, you will see minmum one time a "cell deviation", greater 0.1 Volt for a second or two.
Also I have really doubts, that the height shown in tbe log, after this incidents, is the height, the drone was at.
I had several logs analysed at DJI: Spark was reacting on stick movements
Even when the video showed it banging on the ground, instead of ascending.
I am thinking, that the powet loss, falling 30cm, recovering, goes in the log just as "stick up, moving up"
I just can't prove that, whenever I set up a camera for filming, RC in picture, Spark in picture, the thing works flawlessly. Just last week. 4 batteries no trouble, so started to fly, instead of testing, grabbed the RC for better controll off the "videoboard" and 2 batteries after I had to fight to get it in the air, again.
And the log shows slowy rising heights, to 3 or 4 m higher, as I flew it in the beginning.
And for some time in the flight, it still had powerloss, whoohoo, when trying to ascend.
But in the end it was gone. And: celldeviation, 0,125 Volt for about 2 secs shown in the beginning of the log.

To OP:
Try to start with sticks only, just minimal movement on the left one
I am thinking about setting the stick sensibility of the one a bit down.
If that just wouldn't cut the highspeed, also.

I had a look at a few more log files and the VPS cuts out a few times but this seems normal above 23feet. In the last log which I also linked above it seems that the VPS indeed shows the correct height but the IMU doesn't. I always check what the app tells me about the AC status after booting it up and never have I seen an "not normal" IMU.

I'll analyze a few more log files including the AC black box and look for a pattern. I doubt that DJI will do anything in this regard which is a shame because it's clearly an unusual behavior. I envy the guys who don't have to deal with all of this and can just fly.

I opened another support request because I'm afraid that DJI will charge me if I send in my drone. Keep you posted.

Oh, I did not see you had a log in the topic.
but as I thought, cell deviation



This is, as far as I haved checked, in ALL my "power loosing start attempts"
And I can't recall, to have seen that in any other logfile I have checked, so far.

MartyTravels Posted at 2018-7-13 09:11
I had a look at a few more log files and the VPS cuts out a few times but this seems normal above 23feet. In the last log which I also linked above it seems that the VPS indeed shows the correct height but the IMU doesn't. I always check what the app tells me about the AC status after booting it up and never have I seen an "not normal" IMU.

I'll analyze a few more log files including the AC black box and look for a pattern. I doubt that DJI will do anything in this regard which is a shame because it's clearly an unusual behavior. I envy the guys who don't have to deal with all of this and can just fly.

I don't understand why you think the IMU height is incorrect. It 's value is 0.0 until about 5.5 secs. That's consistent with the zSpeed which is 0.0 until 5.5 secs. It's also consistent with the throttle input which starts being reported at 5.5. secs. The vpsHeight seems to have problems though.


More could be learned about the incident if you could retrieve the .DAT from the Spark. Here are some instructions how to do that.

http://www.datfile.net/DatCon/retrieveV3Dat.html#CompressedFile

Because if I look at the flight log online I see that the VPS height is consistent with the drone going down, while the IMU shows an upward direction.

I'm on the dat file...just need to find the right one. I'll upload it tomorrow.

I'm still checking but if anyone else wants to take a look here is the respective DAT file from the AC

I also exported it in CSV to make it easier :-)

OK: Only the CSV for now. The DAT is too large FLY208.zip (336.65 KB, Down times: 1)

2018-7-13 Upload

Click on the file name to download attach

Here is a separate link to the DAT file DAT file

MartyTravels Posted at 2018-7-13 23:48
I'm still checking but if anyone else wants to take a look here is the respective DAT file from the AC

I also exported it in CSV to make it easier :-)

This is quite the puzzler. I'm still looking at it but thought I would pass along this. There is a discrepancy between the barometer:Raw and barometer:Smooth signals. The smooth value should just be a time averaged version of the raw value. For whatever reason that doesn't appear to be the case for part of this flight.

Additionaly, we can look at the barometer data. The barometer:Raw label in the above is a little misleading. Actually, the real raw barometer data is scaled and biased to yield an altitude. Below shows the real raw barometer data (blue plot)

Hi. I am sorry to hear what happened to your DJI Spark. Since there was an incident of a crash in the Drone I highly suggest to send the Drone in for further assessment and repair. I will be providing you the online repair request link ( www.dji.com/support ) for you to create a ticket for your DJI Spark. Thank you.

Thanks Stephen for taking the effort to write something in regards to this error.

There is one thing that prevented me from opening a repair ticket. If I refuse to repair my spark after DJI quoted a price I might have to pay an assessment fee of 50USD per hour...wow.

Since I doubt that DJI will repair the drone under warranty I want to wait what the support tells me. Well, you guessed it...nobody answered so far :-)

Ok, the bird is flying again. Apparently I got all the sand out of the motors and it's running like before. Now I can collect more data when the error returns.