Mavic Air rapid take off and rapid come back with the crash

PiotrekKa · Apr 8, 2020

Hi All, I'm writing here with hope you can share some light on what's going on as didn't learn anything from DJI response to my question. I had similar situation already twice, let me describe it below.
I did manual take off, then drone rapidly fly approx 10m high hang for a while and come back crashing on the ground. During the whole time drone is not reacting to any attempts to control it. First time it happened in the mountains and now when starting from the building, makes me think it might be somehow related with altitude. Hopefully nothing than propellers got demaged, but I'm little afraid now to start a drone when other people are anywhere around.

sar104 · Apr 8, 2020

Post the flight log.

PiotrekKa · Apr 8, 2020

Please see attached, the latest one

sar104 · Apr 8, 2020

Looks like a faulty IMU. It takes off and climbs because the FC is getting spurious data indicating that it is rapidly descending, and so it applies motor thrust to try to compensate. Then the error reverses, causing it to descend:

This one will need the mobile device DAT file ending FLY072.DAT in order to look at the raw sensor and IMU data.

MrSpok · Apr 9, 2020

@sar104 : Is there any way to get a picture like that with a local app on win10? I don't want to use airdata.
Thanks a lot.

Thomas B · Apr 9, 2020

MrSpok said:
@sar104 : Is there any way to get a picture like that with a local app on win10? I don't want to use airdata.
Thanks a lot.

No

Keule · Apr 9, 2020

MrSpok said:
Is there any way to get a picture like that with a local app on win10?

You might read following link: Flight log retrieval and analysis guide

Then go on and install CsvView on your PC. That can be found here: CsvView/DatCon

slup · Apr 9, 2020

MrSpok said:
@sar104 : Is there any way to get a picture like that with a local app on win10? I don't want to use airdata.
Thanks a lot.

Keule said:
You might read following link: Flight log retrieval and analysis guide

Then go on and install CsvView on your PC. That can be found here: CsvView/DatCon

Thank's to our forum member @BudWalker it is a way ... with, as pointed out, CsvView.

Here's a similar chart as @sar104 made in his Igor Pro software...

MrSpok · Apr 9, 2020

Keule said:
Then go on and install CsvView on your PC.

Many thanks for the link. That's a very interestring post.
I have instaled it, but - unfortunately - it doesn't work. After starting CsvView i get a message that tells me, a 64-Bit Java is neccessary. That leads me to a webside where I can download the required Java version. I'm sure, all at my PC is on newest version. But anyway, I've downloadet and installed it. But no change. So after uninstall both I installed at first Java and then CsvView again. But again, the Java is missing and I always come to the Java webside. It is not possible to get CsvView to run.
Is there any trick to fix that?

Keule · Apr 9, 2020

MrSpok said:
Is there any trick to fix that?

CsvView requires a 64 bit Java to run

Java Downloads for All Operating Systems

Java manual download page. Get the latest version of the Java Runtime Environment (JRE) for Windows, Mac, Solaris, and Linux.

java.com

BudWalker · Apr 9, 2020

MrSpok said:
Many thanks for the link. That's a very interestring post.
I have instaled it, but - unfortunately - it doesn't work. After starting CsvView i get a message that tells me, a 64-Bit Java is neccessary. That leads me to a webside where I can download the required Java version. I'm sure, all at my PC is on newest version. But anyway, I've downloadet and installed it. But no change. So after uninstall both I installed at first Java and then CsvView again. But again, the Java is missing and I always come to the Java webside. It is not possible to get CsvView to run.
Is there any trick to fix that?

Try this. Start a command line interpreter and enter
java -version

Here is what I get when I do this.

MrSpok · Apr 9, 2020

BudWalker said:
Try this.

I did:
"Microsoft Windows [Version 10.0.18362.720]
(c) 2019 Microsoft Corporation. Alle Rechte vorbehalten.

C:\Users\xx>java -version
java version "1.8.0_241"
Java(TM) SE Runtime Environment (build 1.8.0_241-b07)
Java HotSpot(TM) Client VM (build 25.241-b07, mixed mode)

C:\Users\xx>"

There is nothing about 32/64-Bit.
I saw, you have Win 6.x, my Win is 10. Can this be the reason?

Keule · Apr 9, 2020

MrSpok said:
There is nothing about 32/64-Bit.

You have only the 32bit version installed.
It should display this:

java version "1.8.0_241"
Java(TM) SE Runtime Environment (build 1.8.0_241-b07)
Java HotSpot(TM) 64-Bit Server VM (build 25.241-b07, mixed mode)

Install the 64bit version manually from the link given in post #10

PiotrekKa · Apr 9, 2020

sar104 said:
Looks like a faulty IMU. It takes off and climbs because the FC is getting spurious data indicating that it is rapidly descending, and so it applies motor thrust to try to compensate. Then the error reverses, causing it to descend:

View attachment 98247

This one will need the mobile device DAT file ending FLY072.DAT in order to look at the raw sensor and IMU data.

I have downloaded dat file using DJI assistant, the file is from that day. It's quite large so here is the link

DJI_ASSISTANT_EXPORT_FILE_2020-04-09_19-49-19.DAT

drive.google.com

sar104 · Apr 9, 2020

PiotrekKa said:
I have downloaded dat file using DJI assistant, the file is from that day. It's quite large so here is the link

DJI_ASSISTANT_EXPORT_FILE_2020-04-09_19-49-19.DAT

drive.google.com

That's the aircraft DAT file - not readable. It's the mobile device DAT file that is needed here.

PiotrekKa · Apr 9, 2020

It seems I overcomplicated it, please see attached.

slup · Apr 9, 2020

Yeah ... a faulty IMU for sure. A lot of messages in the DAT indicating various IMU related errors ...

Height readings in the TXT vs. DAT is also very different ..?

TXT:

DAT:

sar104 · Apr 9, 2020

PiotrekKa said:
It seems I overcomplicated it, please see attached.

Both IMUs looks wrong, but the active IMU, which was IMU1 (red dashed line), was very wrong - it went down, leading the FC to try to correct, which ended up as the climb that the barometer, IMU0 and the GPS module detected:

Looking at the IMU and GPS vertical velocities:

IMU1 shows spurious downward (+ve) IMU1 vertical velocity (green), leading to actual upward (-ve) velocity (red and black). Integrating with respect to time:

It's clear from these data that IMU1 is in bad shape, but IMU0 is also underestimating vertical motion.

PiotrekKa · Apr 9, 2020

Thank you, I have lost confidence in my pilot skills. Will try to claim it on warranty through DJI.

MrSpok · Apr 9, 2020

@Keule , @BudWalker
many thanks to you for your support. Now it works.

When a software leads me to a webside it is supposed to be the correct webside. Unfortunately this was not the case.
Anyway, I will explore the app the next days.