Mavic Mini 'crash'

[slup]

Wonder if DJI would mail you that DAT file if you ask for it ...

 

[BudWalker]

I am wondering what data DJI has access to.
In the chart they sent the analysis the FLY008.DAT has an index. (and has events from beyond T=977s)

View attachment 98416

Any chance that the OP only has ...FLY008(0) or something like that?

@Croxy how did you submit the .DATs to DJI? And, what Android app was used to retrieve the .DATs that you submitted here? I'm thinking that FLY008(1) could have been ignored. Maybe you could try another method to retrieve the .DATs.

 

[Capt KO]

Wow! When i first read the OP post, I just assumed that he was too low over water. Ending in the common, uncommanded decent into the drink. Seems far different from the usual mini swim. Interesting

Good save and glad you had Care refresh

As asked previously, did all the props survive or was one missing upon retrieval?

 

[Capt KO]

Please read or reread all the suggestions on the forum to keep from crashing your new mini. One of them is to not fly low over water or other reflective services like snow or sand. Often results in an uncommadded decent, especially when hovering.

Fly safe, have fun and try not to use your second warranty.

 

[DanMan32]

Also to note that elevation readings can drift. It may say you're 3 ft up, but actual could be more or less, not even taking into account terrain elevation changes relative to takeoff.

 

[DanMan32]

By the way, you can get Refresh Plus, which gives one additional refresh after your current year expires. Besides the extra refresh, it extends the warranty through to the end of Plus. Regular refresh does not extend the warranty. Many components otherwise have only 6 months.
Not to confuse warranty for manufacturing defects, which doesn't count against you for a refresh, vs an incidental damage replacement.

 

[Capt KO]

By the way, you can get Refresh Plus, which gives one additional refresh after your current year expires. Besides the extra refresh, it extends the warranty through to the end of Plus. Regular refresh does not extend the warranty. Many components otherwise have only 6 months.
Not to confuse warranty for manufacturing defects, which doesn't count against you for a refresh, vs an incidental damage replacement.

Croxy, take note! As stated above, you have the opportunity to upgrade your warranty. Take It, I did.

 

[Croxy]

Yes, all props in good condition and attached as normal (I actually removed them prior to sending the Mavic back to DJI).

@Croxy how did you submit the .DATs to DJI? And, what Android app was used to retrieve the .DATs that you submitted here? I'm thinking that FLY008(1) could have been ignored. Maybe you could try another method to retrieve the .DATs.

Hi Bud,

The Android App used: DJI Fly. After reporting the issue to DJI, I received the following instructions:

1. Click Profile on the home page
2. Please enable the Auto-Sync flight records ( go to Settings -> Sync Flight Data-> Auto-Sync flight records)
3. Then please connect network, the flight records will be uploaded automatically.

I followed those steps, which worked fine. Btw, as I mentioned before, I only made 13 flights in total; all show up in the Flight Data Centre in my phone app. Therefore, perhaps they can be exported again??

Cheers.

 

[slup]

Yes, all props in good condition and attached as normal (I actually removed them prior to sending the Mavic back to DJI).


Hi Bud,

The Android App used: DJI Fly. After reporting the issue to DJI, I received the following instructions:

1. Click Profile on the home page
2. Please enable the Auto-Sync flight records ( go to Settings -> Sync Flight Data-> Auto-Sync flight records)
3. Then please connect network, the flight records will be uploaded automatically.

I followed those steps, which worked fine. Btw, as I mentioned before, I only made 13 flights in total; all show up in the Flight Data Centre in my phone app. Therefore, perhaps they can be exported again??

Cheers.

Unfortunately it's only the TXT logs that can be retrieved from the DJI servers again ... the DAT logs which contain all the raw sensor & motor data (which the TXT doesn't) is lost for you in this case.

But as DJI showed you a chart giving away that they had access to the correct DAT ... so as I proposed in another post, why not ask them to send it to you.

 

[Croxy]

Unfortunately it's only the TXT logs that can be retrieved from the DJI servers again ... the DAT logs which contain all the raw sensor & motor data (which the TXT doesn't) is lost for you in this case.

But as DJI showed you a chart giving away that they had access to the correct DAT ... so as I proposed in another post, why not ask them to send it to you.

Sure, I'll ask the question and post back here if successful. Thanks!

 

[Keule]

@BudWalker @sar104

After digging more into the issue about the .DAT file from this incident I found out:
CsVViewer 3.7.4 has no events after T=997.135
DatCon 3.4.7 also does not produce/converts data point past T=996 in a .CSV

Then I used the data viewer from DJI's Assistant 2 for Matrice and ...

... you see the missing data points.

I have to experiment with other Assistant 2 data viewers if more data fields are accessible.

It looks to me, whatever procedure/function is being used to handle *.DAT files, interprets prematurely EOF and 'forgets' the remaining data points.

 

[BudWalker]

@BudWalker @sar104

After digging more into the issue about the .DAT file from this incident I found out:
CsVViewer 3.7.4 has no events after T=997.135
DatCon 3.4.7 also does not produce/converts data point past T=996 in a .CSV

Then I used the data viewer from DJI's Assistant 2 for Matrice and ...
View attachment 98612
... you see the missing data points.

I have to experiment with other Assistant 2 data viewers if more data fields are accessible.

It looks to me, whatever procedure/function is being used to handle *.DAT files, interprets prematurely EOF and 'forgets' the remaining data points.

Thank you. I can see how DatCon is broken. It's the same problem described here
Automatic Landing Flyaway
Still thinking about the best way to fix it. I'll try to provide a correct .csv in the next day or two.

 

[BudWalker]

@BudWalker @sar104

After digging more into the issue about the .DAT file from this incident I found out:
CsVViewer 3.7.4 has no events after T=997.135
DatCon 3.4.7 also does not produce/converts data point past T=996 in a .CSV

Then I used the data viewer from DJI's Assistant 2 for Matrice and ...
View attachment 98612
... you see the missing data points.

I have to experiment with other Assistant 2 data viewers if more data fields are accessible.

It looks to me, whatever procedure/function is being used to handle *.DAT files, interprets prematurely EOF and 'forgets' the remaining data points.

The .csv is here
20-03-20-09-44-51_FLY008.csv
It's large because it's at the highest sample rate and contains the both the Engineered and DatDefined data. You can submit it directly to CsvView.

The time scale is a bit different from what you're seeing with the DJI Assistant 2 Viewer. That's because CsvView adjusts the clock rate to reflect that this is a Mavic Mini. The Viewer doesn't do this.

I looked at the incident briefly. Doesn't look like lost propulsion to me. Maybe a problem with the gyro itself. This from the eventLog stream
1339.061 : 69879 [L-FDI]NS(0) GYRO(0): fault on , stuck
1339.095 : 69881 [L-RC]craft ctrl failed!!!
:
1339.277 : 69890 [L-FDI][CTRL]: fault off, impact_in_air
:
1339.535 : 69903 [L-FDI]NS(0) GYRO(0): fault on , range
:
1339.638 : 69908 [L-FDI]NS(0) GYRO(0): fault off, stuck
1340.193 : 69936 [L-FDI]fdi_multi_regi_gyr[0] failed: -2

@sar104 , @slup

 

[BudWalker]

And here is the eventLog stream
20-03-20-09-44-51_FLY008.log.txt
@Keule , @sar104 , @slup

 

[slup]

Yeah ... see that indication in the event log. No abnormalities regarding the prop RPM's around 1339sec either ... all signals goes haywire there, AC position, accel & gyro.

I'm not fully confident yet to securely depict prof of HW failures ... what signals can indicate this?

 

[Croxy]

Unfortunately it's only the TXT logs that can be retrieved from the DJI servers again ... the DAT logs which contain all the raw sensor & motor data (which the TXT doesn't) is lost for you in this case.

But as DJI showed you a chart giving away that they had access to the correct DAT ... so as I proposed in another post, why not ask them to send it to you.

Hi again,

I finally received the requested file from DJI technical and it's exactly the same as the one I posted already (20-03-20-09-44-51_FLY008)! However, I get the sense that there is some kind of timing mismatch according to my (probably poor) understanding of Bud's post (#33)??

 

[BudWalker]

Hi again,

I finally received the requested file from DJI technical and it's exactly the same as the one I posted already (20-03-20-09-44-51_FLY008)! However, I get the sense that there is some kind of timing mismatch according to my (probably poor) understanding of Bud's post (#33)??

Actually the incident was included in 20-03-20-09-44-51_FLY008.DAT. Although shrouded in obscure techno-speak post #32 explains this. CsvView/DatCon has a fundamental bug that prevented the incident data from being converted and presented in the .csv. The attachment to post #33 has the incident data for others to look at. The next version of CsvView/DatCon has this bug fixed.

At first glance the incident looks like a propulsion issue, either a failed motor or broken prop. Although spinning around the Z axis there was much less rotation about the other two axes. Normally there would be more or less the same amount of rotation about all 3 axes; i.e, it would be tumbling.

At the onset of the incident (1339 secs) the eventLog shows
1339.061 : 69879 [L-FDI]NS(0) GYRO(0): fault on , stuck
1339.063 : 69879 [L-FDI][IMU] gyro acc 0 err status 4140
1339.077 : 69880 [L-FDI][CTRL]: fault on , tilt_ctrl_fail
1339.095 : 69881 [L-RC]craft ctrl failed!!!

I.E., the gyro is stuck. This can also be seen in the gyro data

where GyroZ was near 0.0 but then sticks at -2069.00. Some, myself included, would say this is physically impossible. Clearly a HW failure.

 

[Croxy]

Thanks Bud, I really appreciate your (and others) input. I'll write to DJI with the findings and let you know what the response is.
Keep safe!
Best wishes..

 

[BudWalker]

Hi again,

I finally received the requested file from DJI technical and it's exactly the same as the one I posted already (20-03-20-09-44-51_FLY008)! However, I get the sense that there is some kind of timing mismatch according to my (probably poor) understanding of Bud's post (#33)??

I forgot to ask? Just to be clear, DJI actually gave you the .DAT after they removed it from your mobile device? That would be HUGE!! Maybe others can get their .DAT from DJI.

 

[Croxy]

I forgot to ask? Just to be clear, DJI actually gave you the .DAT after they removed it from your mobile device? That would be HUGE!! Maybe others can get their .DAT from DJI.

I previously synced my device with the DJI cloud as requested by them initially. Afterwards, I then copied the file and posted it here when I started the topic, so therefore it still existed on my phone.

Following advice here, I then requested the file back from DJI (assuming that the file they had differed from the one I had); it was in fact the same file. However, they did indeed acquiesce to my request for the DAT file.