More Mavic Mini Uncommanded Descents

[rumitcu]

This is my first post here and wanted to begin by thanking everyone for sharing their experiences and analyses.

I have generally had very good experiences with my Mavic Mini but have also experienced several uncommanded descents of my Mavic Mini. I thought I'd share my experiences and see if anyone was willing to take a peak at my log(s) to see if it looks like the same issue that others are reporting.

Here are links to the flights that experienced uncommanded descents:

• March 14, 11:36 am
  • Uncommanded descent into my driveway.
• March 14, 11:33 AM
  • Uncommanded descent into my driveway.
• March 14, 11:13 am
  • Got IMU warning in DJI Fly app, then it descended uncontrollably, and crashed into my deck railing.
  • I did a full IMU calibration and compass calibration before the next flight.
• March 4, 1:12 pm
  • Uncommanded descents into asphalt while flying low (~6' AGL)

Note that I have about 2 dozen successful flights without any issues even though I live at 6800' elevation. These problems seem to be unpredictable--e.g., on March 4, I flew the Mini later that afternoon in the same conditions and didn't have any issues.

After experiencing these problems, I contacted DJI support and they asked me to send the Mini in for inspection/repair. Over the course of two weeks, they inspected the .txt logs from my phone and concluded that I was flying with the propeller guards at high altitude, which is not allowed per the manual, and therefore, the drone was not covered under warranty. However, I did NOT have propeller guards mounted on the drone. Unfortunately, I didn't have a second camera available to video the drone falling from the sky, because then they couldn't claim that the logs were indisputably correct. I argued with them to no end and asked them to send it back to me because it is still functional when it doesn't decide to randomly fall from the sky. Obviously, it was very disappointing to spend $500 on a drone that falls from the sky unexpectedly and without warning.

Any assistance in examining my logs is much appreciated.

All the best,
Pat in Boulder, CO

 

[sar104]

The txt logs are just going to show the effects of reduced thrust - we've seen plenty of those. Are you able to retrieve any of the associated DAT files (FLY024.DAT, FLY030.DAT, FLY032.DAT) - those would be much more useful.

 

[rumitcu]

The MCDatFlightRecords folder was empty on my phone.

 

[sar104]

The MCDatFlightRecords folder was empty on my phone.

If you do experience any more of these events then see if you can snag the DAT files before they get deleted. Alternatively - disable "Auto-sync Flight Records" - that should prevent them from being deleted.

 

[rumitcu]

Will do and thank you.

 

[MAvic_South_Oz]

they inspected the .txt logs from my phone and concluded that I was flying with the propeller guards at high altitude

Did the logs show payload mode on ?

You should get more flights with this happen by the sounds.
I'd fly numerous times in a big open flat grassed area, like a sports ground, and wait to get the TXT / DAT for that flight.

Maybe take some pics of you taking off without prop guards, just as a back up to show DJI you are not using them.

 

[bmwbob]

I have well over 100 flight without any uncommanded descent other than not being very far from the home point. Then I read the manual!

 

[rumitcu]

If you do experience any more of these events then see if you can snag the DAT files before they get deleted. Alternatively - disable "Auto-sync Flight Records" - that should prevent them from being deleted.

Got my Mavic Mini back from DJI, did a quick test flight with no issues (per AirData notifications), and was able to download 2020-04-08_13-01-35_FLY045.DAT, so now I can at least download the .DAT files.

Per Flight log retrieval and analysis guide, I downloaded CsvView to be able to examine the .DAT files for future use, but when I launch the app, I get a message that says, "This application requires a Java Runtime Environment 1.8.0 (64-bit)" despite the fact that I have Java 1.8.0 installed on my PC. Additionally, CsvView and DatCon both state, "Note: The Mavic Air, Mavic 2, and Mavic Mini.DAT will probably be encrypted and can not be processed by either DatCon or CsvView".

So am I barking up the wrong tree trying to be able to read the .DAT files from my Mavic Mini?

Thanks!
Pat

 

[sar104]

Got my Mavic Mini back from DJI, did a quick test flight with no issues (per AirData notifications), and was able to download 2020-04-08_13-01-35_FLY045.DAT, so now I can at least download the .DAT files.

Per Flight log retrieval and analysis guide, I downloaded CsvView to be able to examine the .DAT files for future use, but when I launch the app, I get a message that says, "This application requires a Java Runtime Environment 1.8.0 (64-bit)" despite the fact that I have Java 1.8.0 installed on my PC. Additionally, CsvView and DatCon both state, "Note: The Mavic Air, Mavic 2, and Mavic Mini.DAT will probably be encrypted and can not be processed by either DatCon or CsvView".

So am I barking up the wrong tree trying to be able to read the .DAT files from my Mavic Mini?

Thanks!
Pat

As explained in the guide, the aircraft DAT files from the MA, M2 and MM are not readable, but the mobile device DAT files can still be read. As for Java, you most likely only have the 32-bit version installed.

 

[BudWalker]

......
I downloaded CsvView to be able to examine the .DAT files for future use, but when I launch the app, I get a message that says, "This application requires a Java Runtime Environment 1.8.0 (64-bit)" despite the fact that I have Java 1.8.0 installed on my PC.

As @sar104 mentioned you are most likely running a 32-bit Java version. Lately, more users are running into this problem. I'm looking for ways to mitigate this problem in the next CsvView/DatCon version.

...... Additionally, CsvView and DatCon both state, "Note: The Mavic Air, Mavic 2, and Mavic Mini.DAT will probably be encrypted and can not be processed by either DatCon or CsvView".
....

The CsvView/DatCon website was a bit confusing about which .DAT files are encrypted. The website has been changed to make this more clear.

 

[rumitcu]

Thank you, both! Per @sar104's suggestion, I downloaded a 64-bit version of Java from Java Downloads for All Operating Systems and CsvView started up nominally. Once it started, I was able to confirm that I could open the .DAT file from DJI Fly app and begin to analyze the data. @BudWalker, the statement on the website is definitely more clear now--thanks!

Because the uncommanded descents seem to be an erratic problem that looks similar to flying with prop guards, I did some testing with and without prop guards. And this time, I recorded the flights using the onboard camera, screen recording on my iPhone, and with an external camera pointed at the Mini. The Mini flew reasonably well with the prop guards installed and did not experience any uncommanded descents, although I noticed that the propellers sounded a much higher frequency. Subsequent flights with the prop guards removed confirmed a lower frequency sound for the props, so this may be a key for me to know if the drone is performing properly in the future (although I would prefer an actual warning in the DJI Fly app instead!).

Plotting the motor speeds in CsvView, I can clearly see the higher speeds in the rear propellers compared to the forward propellers when the prop guards were installed that I've heard others speak of in regards to uncommanded descents. Note that the logs also reported several instances of "Not Enough Force/ESC Error" when the prop guards were installed but none when they were removed. Interestingly, the rear motors are canted by ~4 degrees from vertical, so although they may require slightly higher speeds, the effect should be very small (<1%) and wouldn't explain the significant speed difference between the front and rear motors when under heavy loads. I'm baffled but feel much better prepared to analyze the situation if it happens again to me.

 

[rumitcu]

I also just plotted some data from the .txt file from one of my uncontrolled descents, and it's clear that the aircraft is falling from the sky even when full throttle is being applied. What I don't understand is that even if prop guards were installed (payload mode enabled), why would DJI consider this to be acceptable behavior? Clearly the aircraft was able to apply sufficient thrust to maintain a height of 8.5m and then all of a sudden, it just starts to fall from the sky. This is embarrassing and awful.

 

[sar104]

I also just plotted some data from the .txt file from one of my uncontrolled descents, and it's clear that the aircraft is falling from the sky even when full throttle is being applied. What I don't understand is that even if prop guards were installed (payload mode enabled), why would DJI consider this to be acceptable behavior? Clearly the aircraft was able to apply sufficient thrust to maintain a height of 8.5m and then all of a sudden, it just starts to fall from the sky. This is embarrassing and awful.

View attachment 98378

I'm pretty sure this is caused by mechanical instabilities in the props. Having just been sent a Mavic Mini to test by another forum member I'm quite surprised at just how floppy they are. Now I just need to figure out how to reproduce the issue.

 

[rumitcu]

@sar104, that's a good hypothesis. If you're considering structural instabilities, the torsional flexibility of the arms definitely needs to be considered also. When the Mini first came out, one of the most common critiques (for better or worse) was how flimsy the arms are. Perhaps there is some coupling between the motor control, the props, and the arms? This is a challenging problem to debug from a user perspective, whereas DJI certainly has all the tools at their disposal to prove that the aircraft has adequate stability margins from both a control and structural perspective. Just wish they would step up to the plate and address this issue head-on.

 

[Meta4]

DJI certainly has all the tools at their disposal to prove that the aircraft has adequate stability margins from both a control and structural perspective. Just wish they would step up to the plate and address this issue head-on.

DJI's failure to acknowledge the problem or do anything about it is very odd.
It's been some months now and they just keep replacing Minis.
It makes you wonder: is the issue too complex and a solution is just too hard to find or have the wheels have come off at DJI and no-one's doing much at all?

 

[sar104]

DJI's failure to acknowledge the problem or do anything about it is very odd.
It's been some months now and they just keep replacing Minis.
It makes you wonder: is the issue too complex and a solution is just too hard to find or have the wheels have come off at DJI and no-one's doing much at all?

My guess is the fix is not simple, and the cost-benefit analysis favors replacing those that are lost where the owner makes a claim. I suspect that a lot of cases get written off as user error.

 

[slup]

My guess is the fix is not simple, and the cost-benefit analysis favors replacing those that are lost where the owner makes a claim. I suspect that a lot of cases get written off as user error.

My opinion also ... if only a prop problem DJI have had already sprinkled redesigned props all over.

 

[GerdS]

Today I have experienced even two uncommanded descendings in just one flight! Luckily I was able to get altitude again by full throttle and forward speed instead of falling into an acre.

With this bug I cannot trust into the Mini for flying over critical regions, e.g. water.

I have uploaded the log to AirData, the descendings begin at 1m40s and 8m11s.


Here I have uploaded the .log and .txt files: DescendLogs.zip

The .dat file was not available.

 

[Flycaster]

If you are getting the "not enough force error" with the guards on, and no errors like that without the guards, that would lead me to think that the motor/prop/esc combination is not in the sweet spot. Assuming the Payload was switched on,
And add into the flexing of the arms, which, if being brought straight down as fast as possible, it could be getting caught up in it's own propwash, and start a harmonic osccilation with the flexy arms, and that could cause "not enough force" errors, as the esc is trying to control the prop speeds.
Pure guesswork..

 

[slup]

Today I have experienced even two uncommanded descendings in just one flight! Luckily I was able to get altitude again by full throttle and forward speed instead of falling into an acre.

With this bug I cannot trust into the Mini for flying over critical regions, e.g. water.

I have uploaded the log to AirData, the descendings begin at 1m40s and 8m11s.


Here I have uploaded the .log and .txt files: DescendLogs.zip

The .dat file was not available.

Yeah ... shows height instability everywhere, ironic to see how you managed to ascend to 22,5m & come down again nearly completely without using negative throttle. Most of the descents happens without manual input.

Would never trust this drone flying over anything ... note how many red up peaks it is in z-speed (descend) in relation to down going blue peaks which is negative throttle (descend).