Welcome DJI Spark Pilot!
Jump in and join our free Spark community today!
Sign up

First flight, instant crash after takeoff, FLYxxx.DAT?

First of all, thank you very much for the analaysis.

I was waiting for the Spark to gain enough GPS before takeoff - so that's maybe why it is just sitting there idling at 10 watts. I can upload the next DAT file to see if they make more sense, but the "flight time" was just a few seconds, it was powered on and sitting on the ground before that much longer.
Can you see any GPS coords in the DAT file? Cause the flight after the crash was at a different spot so that could give a clue.

I also have a very first DAT file but the date here is strange as it is BEFORE I received the Spark - maybe DJI internal quality test?

However, I will upload the DAT when I am back home (currently at work).
 
Last edited:
First of all, thank you very much for the analaysis.

I was waiting for the Spark to gain enough GPS before takeoff - so that's maybe why it is just sitting there idling at 10 watts. I can upload the next DAT file to see if they make more sense, but the "flight time" was just a few seconds, it was powered on and sitting on the ground before that much longer.
Can you see any GPS coords in the DAT file? Cause the flight after the crash was at a different spot so that could give a clue.

I also have a very first DAT file but the date here is strange as it is BEFORE I received the Spark - maybe DJI internal quality test?

However, I will upload the DAT when I am back home (currently at work).
I don't see any valid GPS coordinate in the file. Raw altitude says 102 meters.
Just post second DAT and I will check it.
 
First of all, thank you very much for the analaysis.

I was waiting for the Spark to gain enough GPS before takeoff - so that's maybe why it is just sitting there idling at 10 watts. I can upload the next DAT file to see if they make more sense, but the "flight time" was just a few seconds, it was powered on and sitting on the ground before that much longer.
Can you see any GPS coords in the DAT file? Cause the flight after the crash was at a different spot so that could give a clue.

I also have a very first DAT file but the date here is strange as it is BEFORE I received the Spark - maybe DJI internal quality test?

However, I will upload the DAT when I am back home (currently at work).
Sorry, I see now that the Home Point was recorded at 650.5 seconds, but it seems it lost GPS quickly.
Longitude 8.52757359
Latitude 49.0041275
Height 30 m
 
Please also upload your TXT flight log here and post a link back here.
 
DJI Flight Log Viewer - PhantomHelp.com

This 100% the flight with the crash. You can also see the "flight" was only 2 seconds and I can see "compass error" note at the end - but not at takeoff! (at the end, the spark was flipping over as it crashed into the wall).

Looking forward to finally learn what happened here!
 
  • Like
Reactions: Manazuma
DJI Flight Log Viewer - PhantomHelp.com

This 100% the flight with the crash. You can also see the "flight" was only 2 seconds and I can see "compass error" note at the end - but not at takeoff! (at the end, the spark was flipping over as it crashed into the wall).

Looking forward to finally learn what happened here!
DJI_ASSISTANT_EXPORT_FILE_8.5.2017 1756.FLY3.DAT is the crash flight.
GPS was ok.
At (1) in red, you started the motors and take off at end of (1). There was only throttle active to get some altitude.
Just before (2) there was some issue with the magnetometer and you initiated a CSC at 155.4 to save the situation and motors stopped at end of (2). One point is that the CSC was really established at 156.0 sec. In between the sticks were requesting the Spark to move and it amplified the movement until the CSC is really applied and motors stop and fall down. In (3) there was compass errors with high magnetometer sensor value (green graph) but this was after it was down.
Capture42.PNG

I see the message of @msinger after I was mid way. @BudWalker is the expert in these kind of issue.
So the question is what happened between (1) and (2). I see that it has gone down from 0.6m at 154.5s to 0.2m at 155.3s and you then initiated a CSC.
When it reached 0.2m, the laser distance sensor (vpsHeight) seems to give again valid data, low like 0.1m where the Spark is then reacting to it by regaining some altitude (154.1s) on its own.
Capture43.PNG
From how was reacting the magnetometer, there was probably some large metallic structure nearby. If the calibration of the compass has some flow it can start to generate issue before the Spark indicate a compass error. But there was something strange at 153.0s with this drop where I don't understand the reason.
Let's see what BudWalker can say about the behavior.
 
Wow thanks a lot guys, really excited about it! After take-off (end of (1)) it immediatelymoved to its left at full speed so I instantly did a CSC (as it was very low above ground, falling down did look like as a much better option then hitting the wall). I think what you describe as movement of the sticks might be me doing the CSC? I learned that with CSC the aircraft will first do what you tell it to do (throttle down, gear and fly backwards/sidewars) before CSC stops the motors?
Cause I am 100% sure that I was only throttling and nothing else when it immediately flew sidewards.

Let's see what @BudWalker can read here.

PS: what about the DJI_ASSISTANT_EXPORT_FILE_4.4.2017 1650.DAT with the crazy date? Is this some quality check by DJI or just wrong date/time? What about GPS coords in this?
 
Wow thanks a lot guys, really excited about it! After take-off (end of (1)) it immediatelymoved to its left at full speed so I instantly did a CSC (as it was very low above ground, falling down did look like as a much better option then hitting the wall). I think what you describe as movement of the sticks might be me doing the CSC? I learned that with CSC the aircraft will first do what you tell it to do (throttle down, gear and fly backwards/sidewars) before CSC stops the motors?
Cause I am 100% sure that I was only throttling and nothing else when it immediately flew sidewards.

Let's see what @BudWalker can read here.

PS: what about the DJI_ASSISTANT_EXPORT_FILE_4.4.2017 1650.DAT with the crazy date? Is this some quality check by DJI or just wrong date/time? What about GPS coords in this?
No coordinate in it, zero sats on GPS.
The firmware was the same and dated June 15 2017, so it should be a short power on where the time was lost. The first GPS lock set it correctly.
All my first series of photos and videos are dated 01.01.1904 with the Spark :)
 
For some reason I didn't get notified even though there were several @BudWalker tags.
Anyway, @msinger in that part of the flight a CSC had been issued, and it was doing it's FORCE_LANDING where there must've been some ferrous material. That's why the magMod went from 1900 to 4200, magYaw/Yaw separated, and a compass error was issued.
upload_2017-8-9_6-32-15.png
upload_2017-8-9_6-32-27.png

@Dronason I just now read your analysis and it looks like the above is describing what happened at your (2).

@Pac the data is somewhat indeterminate but it looks like the launch site had some ferrous material above ground and the Spark flew past it on the way up. That was the cause of the erratic flight you described. At 154.297 the Spark reached an altitude of 0.7 meters and magX and MagY saw a disturbance. This would be consistent with the distorting material being in the same plane as the Spark (i.e. to the side of the Spark). Note also the effect on magYaw/Yaw
upload_2017-8-9_6-32-15.png
upload_2017-8-9_6-32-27.png
upload_2017-8-9_6-32-33.png

CSC was initiated at 155.355 and the motors finally stopped at 157.874. In those 2.4 secs the Spark moved 5.8 meters. I would have expected the Spark to remain stationary. But, it looks like the way it works is that the Spark responds normally to the control inputs until it figures out that it's a CSC. That's what happened here. It pitched up, rolled left, and rotated CW before it determined it was a CSC.
upload_2017-8-9_6-44-47.png
upload_2017-8-9_6-44-53.png

I think this must be a record for the maximum number of plots in a post;)
 

Attachments

  • upload_2017-8-9_5-58-45.png
    upload_2017-8-9_5-58-45.png
    11.9 KB · Views: 10
  • upload_2017-8-9_5-59-2.png
    upload_2017-8-9_5-59-2.png
    14.2 KB · Views: 10
Last edited:
More than likely this is due to the rebar and other metal in the parking area.

The spark compass is very close to the ground. Once you lifted off you broke the influence from the metal and compass started to through errors. Craft disengages from gps mode when this happens.
This is correct. @Jeff7577 this may seem a bit counter intuitive. Lemme explain. Shortly after batteryOn the magnetometer values are used to initialize the Yaw value. After that Yaw is determined mostly from IMU data (gyros and accelerometers). Magnetometer data then has only a small influence on Yaw. If the Spark is turned on in a location that is geoMagnetically distorted the magnetometers will indicate an incorrect heading. Then Yaw will be compromised because it is initialized using the incorrect mag data. After launch when the AC clears the distortion mag data will become correct but the Yaw value won't change because it's dependent on the gyro data. The FC can't know which is correct and will then issue a compass error.

This scenario has happened many times. This from just yesterday.
Compass error

And, this one seems to be popular
Looking for Trouble ??
 
  • Like
Reactions: Blade Strike
This is correct. @Jeff7577 this may seem a bit counter intuitive. Lemme explain. Shortly after batteryOn the magnetometer values are used to initialize the Yaw value. After that Yaw is determined mostly from IMU data (gyros and accelerometers). Magnetometer data then has only a small influence on Yaw. If the Spark is turned on in a location that is geoMagnetically distorted the magnetometers will indicate an incorrect heading. Then Yaw will be compromised because it is initialized using the incorrect mag data. After launch when the AC clears the distortion mag data will become correct but the Yaw value won't change because it's dependent on the gyro data. The FC can't know which is correct and will then issue a compass error.

This scenario has happened many times. This from just yesterday.
Compass error

And, this one seems to be popular
Looking for Trouble ??

If this is true then it’s poor engineering. Can you point me to information showing how an IMU without a compass can accurately calculate heading? I don’t believe it can. Why would the aircraft move in the horizontal plane because of a yaw discrepancy?

I just did a small test on my couch with a tiny magnet while watching the compass interference bar on GO4. A little over 500 causes it to turn red and issue the compass error. Keeping it just below that in the yellow, the indicated aircraft heading was almost 90 degrees off of actual but showed no errors. If the IMU were being used for yaw why isn’t that shown on the bearing indicator instead of compass heading.
 
If this is true then it’s poor engineering. Can you point me to information showing how an IMU without a compass can accurately calculate heading? I don’t believe it can. Why would the aircraft move in the horizontal plane because of a yaw discrepancy?

I just did a small test on my couch with a tiny magnet while watching the compass interference bar on GO4. A little over 500 causes it to turn red and issue the compass error. Keeping it just below that in the yellow, the indicated aircraft heading was almost 90 degrees off of actual but showed no errors. If the IMU were being used for yaw why isn’t that shown on the bearing indicator instead of compass heading.
Sorry, I'm not able to provide you with the rationale behind DJI's methods. I just analyze incidents.. The description that I gave has been verified in literally dozens of incidents. I'd be happy to answer questions about any of these incidents. You could start by looking at the two incidents I referred to in the post right before yours.
 
@Pac the data is somewhat indeterminate but it looks like the launch site had some ferrous material above ground and the Spark flew past it on the way up. That was the cause of the erratic flight you described. At 154.297 the Spark reached an altitude of 0.7 meters and magX and MagY saw a disturbance. This would be consistent with the distorting material being in the same plane as the Spark (i.e. to the side of the Spark). Note also the effect on magYaw/Yaw

CSC was initiated at 155.355 and the motors finally stopped at 157.874. In those 2.4 secs the Spark moved 5.8 meters. I would have expected the Spark to remain stationary. But, it looks like the way it works is that the Spark responds normally to the control inputs until it figures out that it's a CSC. That's what happened here. It pitched up, rolled left, and rotated CW before it determined it was a CSC.

I think this must be a record for the maximum number of plots in a post;)

Thanks BudWalker for finding your way to my crash posting :)

What I really dont understand is that right after takeoff it flew sidewars at a height of 0.1m or 0.15m but not more, it never went higher and it started drifting right after it did loose contact to the ground. There was no compass error shown (however, I didnt check the sensors -> compass value).
So to me, it seems like the issue was rebar. Maybe if I would have kept pushing throttle, it might have gone up high enough to not be confused by the rebar anymore...
 
Thanks BudWalker for finding your way to my crash posting :)

What I really dont understand is that right after takeoff it flew sidewars at a height of 0.1m or 0.15m but not more, it never went higher and it started drifting right after it did loose contact to the ground. There was no compass error shown (however, I didnt check the sensors -> compass value).
So to me, it seems like the issue was rebar. Maybe if I would have kept pushing throttle, it might have gone up high enough to not be confused by the rebar anymore...
I took a closer look at the first 4 secs of the flight, but I can't add much. The problem is that vpsHeight has a resolution of 0.1 meters and just shows 0.0 meters in this interval. The distance measurement shows that the Spark moved 0.1 meters. But it's computed from GPS coords so it's hard to say how how accurate it is. I'd include the plots, but they are just a bunch of straight lines.

Maybe seeing a 0.1 meter drift while at 0.1 meter altitude isn't so unusual. I don't actually know. When I launch my P3 or Mavic I like to get it to 4 or 5 feet as quick as I can in case it decides it wants drift. On a windy day it can move 2 feet while trying to hover in place.

Anyway, it seems the launch site had multiple opportunities for encountering geomagnetic distortions. Definitely one at end of the flight and probably when the Spark passed through 0.7 meters on the way up. Too bad you can't force the Spark into ATTI mode and fly anyway.
 
Stock FW it came with .400 (I know .500 is out but I read about Sparks flying out of the sky so I thought .400 is fine for now) and DJI Go App iOS 4.1.4 (4.1.5 shows me same changelog).

4.1.4 had a lot of bugs so they updated to 4.1.5 to kill all the bugs. I would not fly with 4.1.4.
 
an inexperienced pilot that doesn't know how to fly in ATTI.!

Designs drone for inexperienced noobs. Gives no option to select atti for practice. Switches to atti automatically when everything goes wrong. Blame noob forcrash because didn't practice flying in atti.
 

Members online

No members online now.

Forum statistics

Threads
14,593
Messages
118,799
Members
17,987
Latest member
csdisme