Spark Altitude instability crash


Have just joined the forum to see if there is any useful technical information out there that can explain a recent crash I had.
I have flown over 11 hours of spark time and have been very impressed with it. However a niggling suspiscion that all is not well started to occurr a few months ago. The initial symptons showed up as an unstable altitude readout on my android phone during flightsd. The aircraft remained stable but the altitude would flick from say 20m to 7m erratically . I was not too bothered as I had good control.

recently though when flying in Vietnam on a steep hillside some 25 secs after a normal takeoff the aircraft did an uncomanded descent full up stick was immediately given which initially caused the spark to climb briefly but then descended crashing into a rock. Revcovery showed that no damage had occurre4d except a missing prop ( which was knocked off by the rock).

I have uploaded the flight data and (hopefully) a screen image is attached showing the IMU data and VPS data diverging rapidly

Any knowledgal techy’s out there got a good reason for this crash. I suspect the speed is actually a vertical speed as the spark did not fly far away. DJI service centre in Hong Kong could find no problems but I am not confident as I know I was applying a full up command during the final descent.

thanks for any ideas
ps all systems were normal ,IMU,compass, GPS etc

Please provide a link to the actual flight data such that we can download the .csv files and give a better explanation of the issue. We can’t definatively tell you if the VPS was functioning properly or not without seeing all the data available.

THanks for the quick reply
Here is the link to the complete csv file

I am also giving a link to another flight report file where for some reason the IMU flight altitude drifted from a positive 30ft to a negative 158ft for no apparent reason. This one is a much linger flight

hope these work

I had asked that you supply the link to the uploaded data, as you originally posted the screen shot from. We can get more data from that, than the .csv’s that you linked to Google Drive. However, that said, it does appear that there was some sort of issue with the VPS. The graph below shows your altitudes, both from the IMU ( Barometer based ) and the VPS altitude. Note the intermittency of the VPS. This also shows the full up throttle prior to the end of the data. Without anything else to go on ( Hence why I asked for the direct link to your original upload ) there is not much more available in this file to discern a direct cause, although the VPS is suspect.

As for the second flight, the drop in altitude to the negative value was due to your down stick commands as you can see in the second graph. The value will go negative if you fly below the launch point. This is normal. If you can provide the direct link to your original upload, there may be more data available.

Hi Flydawg

Sorry for not giving the data you require. I have run into a problem uploading to the DJI site when using DJI assistant 2. I have tried to submit a data report but whenever I go to make the submission it always tells me that my mobile phone number is invalid. I have tried all ways of making it work but cannot get an acceptable number, I assume it requires the number?

I can save the files locally…is that any better if I were to link those?

with regard to my second flight submission the aircraft could not have received those down control inputs as it would have crashed so the down control signals do not make sense. the aircraft remained in a stable level as the VPS signal suggests.

To get the csv files the only way was to upload the data.txt files into the dji log viewer then download the csv from there into my google drive and give you a link. They seemed to be pretty comprehensive regarding the data but is there more to see?

perhaps you can put me straight on how to get the dji assistant 2 to upload data files.

thanks again for your help

Just thought maybe these links are what you wanted

Also I was intrigued to see the intermittent VPS signals on the crash flight as I remember the spark descending in a series of dips rather than a continuous descent almost as though when it saw the VPS it tried to recover but then lost contact.

Those are the ones. Yes. I specifically need the Verbose .csv which I will get from there.

Oh great sorry for not understanding

Where do you see the uncommanded descent in your flight log?

Not to leave you hanging. Just wanted to let you know I have been having system issues today. Will look again ASAP.

Still having some computer issues here , but…it looks like this is unrelated to the VPS. ( Granted there still may be an issue with that ) but from this, it has all the indications of a lost prop. Although you did state the props were attached but one was missing. It looks like it was missing prior to the crash and not due to a rock as you stated. At ~23 seconds the aircraft suddenly pitched up, rolled right and began to descend. This is the crash flight. I have not looked in to the other flight in detail as of yet.

As for the second flight. This flight looks normal. You commanded the aircraft below the takeoff point, thusly the negative altitude. This was not not a drift. It was a commanded descent.

Thanks Flydawg for being so helpful.

The flight crash initiated with the spark hitting the side of a large boulder during the uncommanded descent. If you notice the throttle command is full up from 21 secs until , I suspect boulder impact at 23 seconds. I am therefore of the opinion that the prop detatched on impact with the boulder. Having said that though the spark response from 21 secs looks like a climb until the 23 sec point so you may well be right that at that point a prop failure occurred.

With regard to the second flight I cannot explain these large down control imputs or how the spark came to be 158ft below the takeoff position.I am sure I did not make those command signals the terrain would not allow that as it was flat apart from a raised levy that I launched from but that was no more than 20ft high . Is there a possibility of my RC unit putting out false signals? but even so how can the baro altitude have descended to -158ft without the spark crashing .

A quick question is Swaveheight the VPS height?, not sure I understand correctly the units used for this parameter.

Hi again Flydawg

I have spent some time analysing the second flight and by cross correlating data stored on my phone flight report with the recorded flight data onboard it does seem like those down inputs are genuine and I looked at some of the video footage to remind myself of the flight, unfortunately I was not video recording at the point when it went very negative in altitude. However even though I admit I did fly into a negative altitude below the takeoff point it was certainly not as much as -48m. I would estimate at most 15m. Still a mystery to me why the negative altitude got so low.

This has been a very interesting analysis and I feel alot more confident in how to extract useful data after a flight
Thanks very much:grinning: