MA v01.00.0400 Issues (glitch in yaw and slowly turning in yaw)

[brantel]

Problem?
Ever since updating to v01.00.0400 my once flawless MA has started glitching in yaw about 10-20 degrees to the left (very rapid sudden jerk while hovering) about a minute to three after take off. This seems to be 100% repeatable. The glitch does not seem to reoccur after the initial one but I am being very careful at this point and not venturing off to far. Also I have noticed that the yaw is not "locked in" like it used to be. The aircraft slowly turns left or right without any stick input while hovering.

Was unit in a crash?
No

What have you tried so far?
I have re-calibrated the compass and the IMU and the sensor status page shows little to no interference with the sensors or the compass.

What device are you using ( iphone , ipad, Samsung , etc)?
Ipad Pro 10.5"

What firmware are you running ( aircraft, remote controller)?
v01.00.0400

What Go app version are you using?
4.2.14

Any modification? (if so what)
No

Did you change anything or install any apps? (if so what)
No other than update to v01.00.0400


Today's DJI Assistant Drone & Ipad/RC logs here:

Mavic Air Logs - Google Drive

Today's TXT Flight Logs here:

Flight 1 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 042 DAT file. Glitch at 2:22.6
Flight 2 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 043 DAT file. Glitch at 0:48.8
Flight 3 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 044 DAT file. Glitch at 0:44.3
Flight 4 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 045 DAT file. Glitch at 1:15.7

 

[brantel]

I have just reloaded the firmware with DJI Assistant. Will fly again tomorrow and see if it helped...

 

[sar104]

There are no obvious issues in the first log file. If there are compass issues causing the problem then will need the magnetometer data from the mobile device DAT file pin it down.

 

[ChrisW81]

I’ve just updated my MA firmware and haven’t noticed any adverse movements. Keep us posted.

 

[Darren22]

My Mavic Air is doing the same. Also noticed getting some motor warnings I’ve never had before.

 

[brantel]

There are no obvious issues in the first log file. If there are compass issues causing the problem then will need the magnetometer data from the mobile device DAT file pin it down.

In the original post I included links all the logs that I am aware of. If there are some that I did not include, please let me know how to get them and I will post them.

Here is my link:

Mavic Air Logs - Google Drive

 

[sar104]

In the original post I included links all the logs that I am aware of. If there are some that I did not include, please let me know how to get them and I will post them.

Here is my link:

Mavic Air Logs - Google Drive

Yes - I missed those. The DATs are in there.

 

[brantel]

I have added descriptions to my original post where I have been digesting the log data.

I also wanted to post these graphical images of the log data to show you what I am seeing on each flight (speaking to the glitch).

I have not determined if the slow rotation in hover is the drone or the gimbal moving.

Flight One (Logs in original post)


Flight Two (Logs in original post)


Flight Three (Logs in original post)


Flight Four (Logs in original post)

 

[brantel]

Yes - I missed those. The DATs are in there.

Thanks for the help!

I would love to know how to dig into those DAT files for dissection. Any hints?

 

[sar104]

I'm having trouble matching your flight records but below is FLY045, which shows one of the glitches that you mentioned:



That appears to be the least of the problems here - there are significant heading errors from disagreement between the IMU yaw and the magnetic yaw. The slow, uncommanded turn is apparent too, but with no indication of why it is turning - the FC does not appear to be trying to fix a magnetic error, and in fact is making it worse.

I suggest a full IMU calibration and check the compass calibration.

 

[brantel]

I'm having trouble matching your flight records but below is FLY045, which shows one of the glitches that you mentioned:

That appears to be the least of the problems here - there are significant heading errors from disagreement between the IMU yaw and the magnetic yaw. The slow, uncommanded turn is apparent too, but with no indication of why it is turning - the FC does not appear to be trying to fix a magnetic error, and in fact is making it worse.

I suggest a full IMU calibration and check the compass calibration.

Thanks! I finally figured out how to use CSVView to open the DAT files from the controller/tablet. Seems like the DAT files from the drone are encrypted and can't be opened by CSVView. Also one of my DAT files (FLY043) can't be opened by CSVView for some reason.

The images I posted of the glitches above are pulled from data in the .TXT files. There are some difference in them vs the DAT data. I assume the DAT data should be preferred over the TXT data?

I will do a cold IMU calibration, recalibrate the compass and fly again and report back. Last night I already reloaded the firmware using DJI assistant.

 

[sar104]

Thanks! I finally figured out how to use CSVView to open the DAT files from the controller/tablet. Seems like the DAT files from the drone are encrypted and can't be opened by CSVView. Also one of my DAT files (FLY043) can't be opened by CSVView for some reason.

The images I posted of the glitches above are pulled from data in the .TXT files. There are some difference in them vs the DAT data. I assume the DAT data should be preferred over the TXT data?

I will do a cold IMU calibration, recalibrate the compass and fly again and report back. Last night I already reloaded the firmware using DJI assistant.

I also found that the latest version of DatCon would not open FLY043, but previous versions did. Two of the images that you posted appeared to correspond with 44 and 45, but I could not reconcile them with the txt logs. Also, the DAT file conversions seemed to be missing the RC stick data in the engineered fields, and while they were in the DAT-defined fields the data did not make much sense.

Might need @BudWalker to check out those issues.

 

[brantel]

I think I had some of my images and times mixed up. I will fix the prior postings.

Flight 1 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 042 DAT file. Glitch at 2:22.6
Flight 2 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 043 DAT file. Glitch at 0:48.8
Flight 3 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 044 DAT file. Glitch at 0:44.3
Flight 4 is DJI Flight Log Viewer - PhantomHelp.com which is associated with 045 DAT file. Glitch at 1:15.7

Hope this helps clarify.

 

[BudWalker]

..... Also one of my DAT files (FLY043) can't be opened by CSVView for some reason.
....

FLY043 has a corrupted DatHeader. You can get by this by going to the menu bar and selecting DatCon Parsing Options -> Invalid DatHeader OK.

 

[sar104]

FLY043 has a corrupted DatHeader. You can get by this by going to the menu bar and selecting DatCon Parsing Options -> Invalid DatHeader OK.

That doesn't work for me in 3.5.0. It won't load the file at all and reports "Can't Go: .DAT file not specified, lower is greater than upper".

3.4.0 opens it just fine.

 

[brantel]

FLY043 has a corrupted DatHeader. You can get by this by going to the menu bar and selecting DatCon Parsing Options -> Invalid DatHeader OK.

Thank you, I was able open it using that option.

 

[brantel]

Ok, learned a bunch today on how to get the right logs and decode them.

There seems to definitely be something wrong with my drone. Either the latest firmware killed it or something is failing.

Since the prior mentioned flights, I have resinstalled the firmware with DJI Assistant, recalibrated the IMU, the compass and the gimbal. These task did not help!

I then flew 3 more flights this evening and here are the logs:
DJI Flight Log Viewer - PhantomHelp.com
DJI Flight Log Viewer - PhantomHelp.com
DJI Flight Log Viewer - PhantomHelp.com

Here is a link to the .txt and .dat files off of the tablet:
Mavic Air Saga Day 2 - Google Drive

I experienced different kinds of glitches on all three of these flights. Some yaw glitches like the ones described on the post above, some general glitches where the drone and the motors just make a sudden rapid glitchy sound and the drone shakes a bit, and the issue with the drone slowly turning in a hover. I tried all kinds of movements with the rudder stick to gather different data.
At a few times I let the drone slowly rotate just to see how far it would go and I noticed it rotated so far one direction then started rotating back in the opposite direction.

 

[brantel]

Here is a pic typical of what is weird about the yaw being reported by both IMU's. I would expect these to almost overlay each other but instead there is some significant error between the 2 and I have no idea what is going on with IMU(0) where it is drifting:

 

[BudWalker]

I've noticed a few things about the incident that took place in FLY045. Maybe this will get us a little closer to figuring out what happened.

IMU:ATTI(1) is the correct IMU. That matches the .txt file. And, it makes more sense than the IMU:ATTI(0).

Here's the "glitch" reported by @brantel

But, it's not the Yaw (blue line) which is about .4 secs wide. Rather, it's the magYaw (red line) which is about 2 secs wide. This matches @brantel 's description of having rotated left (CCW) 15° to 20°. This also matches the totalGyroZ.

Using the highest possible Sample Rate and looking closer it appears that the rotation is the result of a glitch in the Yaw value. I.e. the actual rotation starts about .15 secs after the change in Yaw. Too bad we can't look at the on board .DAT to confirm this.


This looks much different than the IMU:ATTI(0) data.


There was another smaller glitch earlier at time 57.33 that may not have been noticed.

 

[BudWalker]

That doesn't work for me in 3.5.0. It won't load the file at all and reports "Can't Go: .DAT file not specified, lower is greater than upper".

3.4.0 opens it just fine.

I'm not sure why this is happening to you. It looks good from here. But, I'm supposing it's somehow related to the stuff that was added to 3.5.0 that persists various settings. The problem occurs with 2018-05-10_19-58-48_FLY043.DAT ? And, does it make any difference if you've selected Smart Time Axis processing?


Please provide the .dotCon file that's in your home directory.