A Legitimate Near-Flyaway with My Fourth M2Z

[Kazzaaz]

I lost two M2Zs back before I took flying seriously.
The third I lost in an apartment fire last month.
This week I bought my fourth with the insurance check and almost lost that one too.

The weather was pretty bad but I was pretty eager to take the drone out for a test flight. Really was just planning on circling my hotel since it wasn't the best area to fly.
As it was over the building I realized the altimeter was incredibly off - the building couldn't have been more than 200ft tall, but according to the AC it was already 500ft!
By the time I realized what was going on, the drone had made its way over the street and began to descend rapidly to maintain what it thought was a steady altitude.

Luckily I maintained VLOS and landed safely. The altimeter read 800ft on the ground at the same spot I took off from!
If I had not raised the altitude limit mid-flight, the M2Z would have flown into the ground.

After the flight is when I receive the system warning - "Main controller data error" and "Barometer Error"
I rebooted, reflashed firmware, and calibrated IMU several times to no avail.

Sometimes when I power on the AC the altimeter drops instead of raising. I am lucky during my flight the barometer went the other way!
If the latter had happened, it would have likely flown on its own hundreds of feet into the air beyond RC control and recovery would be near impossible.
So this bird's next flight will be a one-way trip to DJI warranty department.

Here is a log from two flights, just for the analysts out there. The first flight is what's described above.
The second is a "flight" where really the motors were idling on the ground - not 80ft in the air as shown.
Towards the end it began to slightly pick itself up off the ground (despite full throttle down) and I thought this is it - this is an actual flyaway!
I actually could not stop the motors and had to (nervously) power off via the battery. I still have no idea how I didn't lose a finger.

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

This reminds me of one particular video of an Inspire ascending on its own into commercial airspace. I was always skeptical of that video and thought something else had to be at play.
But although this forum is rightfully cynical, these "flyaways" could and apparently do happen from time to time.

Has anyone else had a similar experience?

 

[Prismatic]

? I'll be an interested bystander.

 

[new2mavic]

In the second log file the VPS was not reporting altitude at all making the drone think that it has lift off

 

[sar104]

For the initial part of the first flight the aircraft climbed and descended appropriately for the throttle inputs that you provided and at exactly the expected rates.



You initially applied full throttle for 50 seconds, and the nominal specified climb rate of 4 m/s was mostly achieved apart from a short period where the wind speed caused sufficient tilt that it had to reduce the rate and throw a "not enough force" error. As a result it ended up a bit short of the 200 m (656 ft) that would have been expected.



After that you applied several periods of full throttle that took it to 210 m.

The barometric altitude does appear to go wrong after that. RTH is followed by autoland, and at 180 seconds the VPS height re-engages at 9.6 m, but the IMU is reading 236 m.



That needs the mobile device DAT file (FLY003.DAT) in order to take a look at the raw sensor data.

How to retrieve a mobile device DAT file: How to retrieve a V3.DAT from the tablet

Flight log retrieval and analysis guide

 

[MercHornet]

Just wondering if you took off from a balcony of your hotel, If so the metal in the floor could affect your drone .

 

[slup]

...You initially applied full throttle for 50 seconds, and the nominal specified climb rate of 4 m/s was mostly achieved apart from a short period where the wind speed caused sufficient tilt that it had to reduce the rate and throw a "not enough force" error. As a result it ended up a bit short of the 200 m (656 ft) that would have been expected.

This is perfectly logical with full throttle during 50sec & were it then ends up altitude wise ... but can't match this with the OP's story & info about altitude.
Hmm ... strange, either something is wrong with the throttle signal also or the OP severely misjudged the altitude.

To be continued ... as soon the DAT is available ?

 

[sar104]

This is perfectly logical with full throttle during 50sec & were it then ends up altitude wise ... but can't match this with the OP's story & info about altitude.
Hmm ... strange, either something is wrong with the throttle signal also or the OP severely misjudged the altitude.

To be continued ... as soon the DAT is available ?

It looks to me as if the OP misjudged the altitude, since the sticks, vertical velocity, barometric and VPS altitude all agree during the initial climb. But that doesn't explain the lack of descent on the barometric altitude.

 

[slup]

It looks to me as if the OP misjudged the altitude, since the sticks, vertical velocity, barometric and VPS altitude all agree during the initial climb. But that doesn't explain the lack of descent on the barometric altitude.

Yeah ... agree, it's something more fishy here

 

[BudWalker]

There seemed to be problems with the barometric pressure earlier in the flight also. There were 2 intervals where VelZ > 0.0 indicating the M2 was descending. But, the height, derived from the barometer, showed the M2 either ascending or remaining at the same height.


At least qualitatively this could match the OP's description.

 

[sar104]

There seemed to be problems with the barometric pressure earlier in the flight also. There were 2 intervals where VelZ > 0.0 indicating the M2 was descending. But, the height, derived from the barometer, showed the M2 either ascending or remaining at the same height.
View attachment 90600

At least qualitatively this could match the OP's description.

Good point. If you compare the time-integrated vertical velocity data to the barometric altitude it starts off fine, but then starts to disagree.



But the integrated velocity data still don't bring it back anywhere near ground level.

 

[Kazzaaz]

I took off from the parking lot, not the balcony as one user mentioned.
Even the integrated velocity curve doesn't look at all right. Both the integrated velocity and the barometric height should have read zero at the beginning and end of the flight.
Maybe both the IMU and barometer are whack?

I never had big intentions with that flight but I will say the wind that day was very strong so it may have introduced quite a bit of inaccuracy.
I also believe the ascent rate at the beginning was slower than 4m/s - I felt that something was wrong before I raised the 400ft height limit (to maintain VLOS over a building that was not 400ft!). Yes, I was pushing full throttle while doing so.
I didn't realize the throttle commands at the end were as erratic as they appear in the graph, but keep in mind the entire end of the flight after RTH was actually closer to what I'd estimate around 150-200ft as I was struggling to keep VLOS and figure out what was wrong.

Something is clearly wrong with this drone, not just this flight. I am just curious if this is a bad barometer, or something deeper.
Upon startup, when completely stationary on the floor, the altitude drops quickly at a steady rate of around 10ft/s (can verify exact rate later) until an error is detected (usually when the altimeter reads around -80ft) and takeoff is prevented.
When I am back to my hotel tonight/tomorrow I will post a video of this strange behavior.

Let me know if there is anything else that I can do that may be of interest to the invested parties here before I send it out!

 

[sar104]

I took off from the parking lot, not the balcony as one user mentioned.
Even the integrated velocity curve doesn't look at all right. Both the integrated velocity and the barometric height should have read zero at the beginning and end of the flight.
Maybe both the IMU and barometer are whack?

I never had big intentions with that flight but I will say the wind that day was very strong so it may have introduced quite a bit of inaccuracy.
I also believe the ascent rate at the beginning was slower than 4m/s - I felt that something was wrong before I raised the 400ft height limit (to maintain VLOS over a building that was not 400ft!). Yes, I was pushing full throttle while doing so.
I didn't realize the throttle commands at the end were as erratic as they appear in the graph, but keep in mind the entire end of the flight after RTH was actually closer to what I'd estimate around 150-200ft as I was struggling to keep VLOS and figure out what was wrong.

Something is clearly wrong with this drone, not just this flight.
Upon startup, when completely stationary on the floor, the altitude drops quickly at a steady rate of around 10ft/s (can verify exact rate later) until an error is detected (usually when the altimeter reads around -80ft) and takeoff is prevented.
When I am back to my hotel tonight/tomorrow I will post a video of this strange behavior.

Let me know if there is anything else that I can do that may be of interest to the invested parties here before I send it out!

As I mentioned above - further investigation is going to require the mobile device DAT file.

 

[slup]

If nothing besides what you have described have happened to the drone & If stationary on the floor and the altitude drops I don't think you will be having any problems with DJI ... I would go ahead as all that's needed for further investigation are in your phone.

For educational purposes the DAT file would be valuable ... perhaps our wisest guys can pinpoint the reason.

 

[Kazzaaz]

Sadly the last recorded .DAT files are from October, with my last drone.

 

[Kazzaaz]

When I am back to my hotel I will clear out the DAT files and try to reinstall GO4.
I'll try one more 15 second flight and hopefully the .DAT is recorded and we are able to enlighten ourselves...
But I can't risk the bird deciding to fly into the clouds to "maintain height", so I will be tying a string to it this time.?

 

[BudWalker]

I had a look at the 2nd .txt where the M2 never left the ground. Both, height and the velZ show the M2 ascending.


I'm starting to suspect that it's the Z axis accelerometer that is a problem. Maybe height is a fusion product with velZ at a higher gain than the barometer. In both this and the previous .txt there is a hard reset of the height at the very end. This would be consistent with height mostly following velZ with correction by the barometer, and possibly, VPS.

@Kazzaaz for your test can you take the props off, start the motors, and then let sit for 3 minutes.

 

[Peio64270]

Yes, I had a similar problem problem with my Mavic Pro 1 (about three years ago, now). Altitude was displayed as decreasing rapidly after take off. After 2 minutes of flight (or so) it could reach negative values as low as -90m AGL while I was hovering or flying slowly at +20m and 20m distance from me. These negative values were still displayed after landing manually at the very same place of take off. The difference with your case is that the drone was not ascending nor descending and that I had no error message in the app (Litchi or DJI Go). Strangely, it kept its true altitude perfectly despite this wrong data and commanded changes in altitude were taken into account though descents were always measured exagerated (10m down instead of observed 5m, for example). Since the drone was under warranty (1 month old) I opened a case to DJI who sent me a new Mavic (not the same SN) saying that they had repaired the first one by changing the motherboard. I still use this second MP which works perfectly for its third anniversary, next month

 

[Kazzaaz]

No matter what I do I cannot get the remote .DAT files to record.

Weirdly enough it will let me take off indoors but not outdoors. My thoughts are, if the accelerator is faulty as BudWalker suspects, pressure may be stable indoors enough for the AC to rely on the barometer alone - but not outside in the wind.
I did another IMU calibration just for the hell of it and without the props, indoors, everything seemed fine.

So I took it outside again. Same result, except this time it actually let itself smack into the pavement. Not that worried since I already have the ticket open with DJI for replacement. I started recording just in time for the plummet which you can compare to the flight data...

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

I really wish I could get a .DAT for you guys. I reinstalled the entire app and cleared all the caches in the root folder so I am out of ideas.

 

[BudWalker]

No matter what I do I cannot get the remote .DAT files to record.

Weirdly enough it will let me take off indoors but not outdoors. My thoughts are, if the accelerator is faulty as BudWalker suspects, pressure may be stable indoors enough for the AC to rely on the barometer alone - but not outside in the wind.
I did another IMU calibration just for the hell of it and without the props, indoors, everything seemed fine.

So I took it outside again. Same result, except this time it actually let itself smack into the pavement. Not that worried since I already have the ticket open with DJI for replacement. I started recording just in time for the plummet which you can compare to the flight data...

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

DJI Flight Log Viewer | Phantom Help

Upload and view flight logs from your DJI drones

www.phantomhelp.com

I really wish I could get a .DAT for you guys. I reinstalled the entire app and cleared all the caches in the root folder so I am out of ideas.

The .txt would provide some info if you can post that.

 

[Kazzaaz]

The .txt would provide some info if you can post that.