Mavic suddenly switched to ATTI mode mid flight

[ankalagon]

Hello everyone. I am new to piloting drones and two days ago I had a pretty scary incident with my Mavic Pro. My Mavic suddenly switched to ATTI mode in the middle of the flight. Until now it flown without any issues, and I was able to fly for about 24 kilomiters (4 hours in the air) perfectly fine. But two days ago it suddenly displayed two warnings (one right after another):

1. Compass Error. Exit P-GPS Mode
2. [IMU]heading exception (In Flight) pleas switch to atti mode if craft behave abnormally (Using)

Then it switched into ATTI mode and started drifting in the wind. About 30 seconds later when I was trying to return the drone home and land suddenly everything started working correctly again.

Everything ended well, but it was a really scary experience. Does anyone had a similar problem? Is it something "normal" (by normal I mean something that can happen sometimes for example when flying into some magnetic field or something)? In that place where I was flying where only fields and few small water ponds. Nothing special that could generate electromagnetic fields or other interference.

I never did any calibrations on my Mavic since DJI Go App never prompted me to do that. Should I calibrate compass or IMU to minimize the chance of this error appearing again?

Right now it looks like my Mavic is working correctly. Also before every flight I try to check if compass is ok. I do this after takeoff by aligning the drone to the object that is visible on the map and rotating the drone to see if the arrow on the map rotates as expected. Right now compass looks ok, but before that flight it also looked like its working correctly.

I have uploaded DAT file logs to my OneDrive folder. I wasn't really sure which one of this files is the one from that flight so I've uploaded both. I think that FLY080.DAT is the one:

2017-07-20 Records

Please let me know if there's anything wrong with my Mavic or is there any maintenance I should do to minimize the risk of it switching to ATTI mode mid flight?

 

[iPa64]

Hi,
Exactly same behavior for me !
16 gps suddenly switch to ATTI mode
But had to landing then switch all off, startup all again and all was ok.

1. Compass Error. Exit P-GPS Mode
2. [IMU]heading exception (In Flight) please switch to atti mode if craft behave abnormally (Using)

 

[Kilrah]

Happens, do a good compass calibration

 

[Cyberpower678]

With an IMU error, I would recommend an IMU calibration. Please read the manual on how to do it correctly.

 

[Kilrah]

The "IMU heading error" is only a consequence of the "compass error" that happened just before.

 

[BudWalker]

Hello everyone. I am new to piloting drones and two days ago I had a pretty scary incident with my Mavic Pro. My Mavic suddenly switched to ATTI mode in the middle of the flight. Until now it flown without any issues, and I was able to fly for about 24 kilomiters (4 hours in the air) perfectly fine. But two days ago it suddenly displayed two warnings (one right after another):

1. Compass Error. Exit P-GPS Mode
2. [IMU]heading exception (In Flight) pleas switch to atti mode if craft behave abnormally (Using)

Then it switched into ATTI mode and started drifting in the wind. About 30 seconds later when I was trying to return the drone home and land suddenly everything started working correctly again.

Everything ended well, but it was a really scary experience. Does anyone had a similar problem? Is it something "normal" (by normal I mean something that can happen sometimes for example when flying into some magnetic field or something)? In that place where I was flying where only fields and few small water ponds. Nothing special that could generate electromagnetic fields or other interference.

I never did any calibrations on my Mavic since DJI Go App never prompted me to do that. Should I calibrate compass or IMU to minimize the chance of this error appearing again?

Right now it looks like my Mavic is working correctly. Also before every flight I try to check if compass is ok. I do this after takeoff by aligning the drone to the object that is visible on the map and rotating the drone to see if the arrow on the map rotates as expected. Right now compass looks ok, but before that flight it also looked like its working correctly.

I have uploaded DAT file logs to my OneDrive folder. I wasn't really sure which one of this files is the one from that flight so I've uploaded both. I think that FLY080.DAT is the one:

2017-07-20 Records

Please let me know if there's anything wrong with my Mavic or is there any maintenance I should do to minimize the risk of it switching to ATTI mode mid flight?

I took a look at this flight. I'm of the opinion that a compass calibration issue is almost never the cause of an incident. I've never seen an incident where there was data or some compelling reason to conclude that a bad compass calibration was the cause. But, the data for this incident is somewhat consistent with what would be expected with a bad compass calibration. Could be other things as well. Not really a question of is the compass calibration bad, rather could it be enough better that this incident wouldn't have happened. If you could perform an experiment maybe we can find out.

At time 122 magYaw (the heading determined by the compass) and Yaw (heading determined by the FC using mostly IMU info) had become 22 degrees different. This then caused a YAW_ERROR_LARGE error and the switch to ATTI. It also caused gpsHealth to drop to 0.


With a bad compass calibration it's expected that magYaw/Yaw separations are heading dependent. That seems to be happening with the early part of this flight.

It's not as pronounced as the data for experiments that I've done with my P3. But, the purpose of those experiments was to create a bad compass calibration and measure the effects.

What then caused the magYaw/Yaw separation? magYaw values become less precise as the geoInclination approaches +90 or -90 degrees. This flight was at 70 degrees. It's also worth noting that if the Mavic had been using compass #2 (front compass) there wouldn't have been the magYaw/Yaw separation, and no incident.

Is there any extra equipment attached to the Mavic; e.g. a tracker?

Are you interested in performing an experiment? Two flights with a compass calibration in between. After launch ascend to 20 meters or so, then rotate the Mavic 2 or 3 times at about 10 secs/360 degrees, then return and land. Do the compass calibration. Then do another flight just like the first one. The .DAT for the two flights will tell us if the compass calibration had any effect

 

[Scottd9000]

It's unfortunate that DJI doesn't allow us to flip on atti mode manually like the older drones. Back in P1 days gps was glitchy at times and it was a good idea to learn to fly in atti mode. Home lock saved my drone a couple time too (switch it on and pull back on right stick it comes back to you)

 

[JRSanta]

Home lock works well when compass isnt working well?

 

[ankalagon]

Are you interested in performing an experiment?
​

I would love to do the experiment and I will definitely try. Unfortunately the results might not be very helpful because I've already did one compass calibration yesterday. I was checking the settings of the DJI Go App and saw that interference of the compass 1 was much greater than those from compass 2 (compass 1 interference between 20-100 with avarage about 35, compass 2 between 0-20 with avarage about 7).

I do plan doing another calibration becouse I think that the last one I made was done pretty poorly. Since it was my first calibration my hands where a bit shaky and place was not perfect (to close to a building). In the result interference in compass 1 stayed the same (or maybe they even increased slightly). After watching almost every tutorial on youtube, on how to calibrate compass correctly, I intende to try one more time this time in more open area.

@iPa64 have you tried any calibrations after your incident? If not maybe you would like to do the experiment? Your data would probably be more reliable then mine now.

 

[BudWalker]

Are you interested in performing an experiment?
​

I would love to do the experiment and I will definitely try. Unfortunately the results might not be very helpful because I've already did one compass calibration yesterday. I was checking the settings of the DJI Go App and saw that interference of the compass 1 was much greater than those from compass 2 (compass 1 interference between 20-100 with avarage about 35, compass 2 between 0-20 with avarage about 7).

I do plan doing another calibration becouse I think that the last one I made was done pretty poorly. Since it was my first calibration my hands where a bit shaky and place was not perfect (to close to a building). In the result interference in compass 1 stayed the same (or maybe they even increased slightly). After watching almost every tutorial on youtube, on how to calibrate compass correctly, I intende to try one more time this time in more open area.

@iPa64 have you tried any calibrations after your incident? If not maybe you would like to do the experiment? Your data would probably be more reliable then mine now.

We already have data from before the calibration; i.e., FLY080. So, now just do the test flight If you do another calibration then do a second test flight after that.

I'm not surprised that there wasn't much change. Shaky hands don't have any effect. In fact, I once did the calibration on my P3 upside down, both the horizontal sweep and the vertical rotation. The requirement to calibrate away from metal is way overstated. All that's really required is that the geomagnetic field be straight within the confines of the compass dance.

BTW, you don't have to actually fly it. You can do three rotations while holding the Mavic in front of you.

 

[ankalagon]

Finally the weather got better and it stopped raining so I was able to do some flying . I have uploaded DAT files from today (FLY86.DAT and FLY087.DAT). I don't understand anything from those charts so I don't have any idea which one is which. When I was driving to the place where I wanted to do the experiment I saw beautiful sunflower field and just had to film it . So probably there might be some logs from this flight somewhere at the beginning. As for the experiment I have placed the drone 20 meters above the ground and rotated it in place. Then I've landed and did compass calibration. Then I have repeated first step (rotating the drone at 20m altitude).

I hope I was able to do the calibration correctly. There was a house about 12 meters from my location but it was a very old wooden house (no concrete). Other than dat only grass and some trees. Please let me know how those readings look now and is my compass calibrated ok?

The link to DAT files is the same: 2017-07-20 Records

PS: I forgot to answer in previous post: I don't have any equipment attached to my Mavic (no GPS tracker or anything similar). I thought about buying extended landing gear but decided to go with the landing pad instead. So in the air there is only Mavic without anything attached.

 

[BudWalker]

Finally the weather got better and it stopped raining so I was able to do some flying . I have uploaded DAT files from today (FLY86.DAT and FLY087.DAT). I don't understand anything from those charts so I don't have any idea which one is which. When I was driving to the place where I wanted to do the experiment I saw beautiful sunflower field and just had to film it . So probably there might be some logs from this flight somewhere at the beginning. As for the experiment I have placed the drone 20 meters above the ground and rotated it in place. Then I've landed and did compass calibration. Then I have repeated first step (rotating the drone at 20m altitude).

I hope I was able to do the calibration correctly. There was a house about 12 meters from my location but it was a very old wooden house (no concrete). Other than dat only grass and some trees. Please let me know how those readings look now and is my compass calibrated ok?

The link to DAT files is the same: 2017-07-20 Records

PS: I forgot to answer in previous post: I don't have any equipment attached to my Mavic (no GPS tracker or anything similar). I thought about buying extended landing gear but decided to go with the landing pad instead. So in the air there is only Mavic without anything attached.

I took a look. In summary, there is a good possibility that the compass cal you did after the incident flight fixed a problem. It's not entirely clear though.

FLY086 is the .DAT where you did two test flights with a compass cal in between. The two test flights show magYaw/Yaw behavior that doesn't get any better. So the compass cal didn't have much effect because there was no improvement to be made.

Comparing the first flight in FLY086 against the incident flight in FLY080 there is a definite improvement in magYaw/Yaw behavior. The fact that you did a compass cal between these two flights makes it likely that that compass cal fixed a problem that caused the incident. This is the first time I've seen some data that supports the claim that a bad compass cal caused an incident.

I took another closer look at the magnetometer data in FLY080. In particular, magY seems to be the culprit that causes magYaw errors. There is an interval where the change in heading is monotonic but magY doesn't follow like it should. The marked areas should be smoother; more like a sine wave.

It should look closer to FLY086 which occurred under ideal conditions


It's not entirely clear though. magYaw has to be corrected for pitch and roll and this becomes error prone when the geoInclination approaches 90 degrees. FLY080 was performed 250 KM north of FLY086's location. I would tend to discount this though since the change in geoMagnetic inclination was only 2.5 degrees. Another consideration is that FLY080 was a real flight with lots of pitch and roll whereas FLY086 has very little.

Got all that?. My advice is to do a "real" flight and then we'll look at the .DAT.

 

[ankalagon]

Thanks for all the info. It's a lot to process but every time I read it it's getting clearer . As for the real flight - today weather got worse but as soon as it's get better I'll try to do some more realistic tests. But for now I think I do have another log that might contain some information. I have added FLY085.DAT file to the OneDrive folder. It should contain the log from that sunflower field I mentioned earlier. I took 180 degree turn there, while flying forward (maybe a bit similar to the turn from the incident flight). I'll upload more logs from real flights as soon as possible.

PS: As for the distance from the incident flight location - it happened while I was on holiday, quite far from home. The best test results would probably require flying in similar conditions in the same location. Unfortunately I won't be able to visiting that place anytime soon. But probably this fall (first half of september) I should be visiting place a lot closer to that location. And even if not, I'll sure be able to fly the drone in place at least 230 km to the north.

 

[BudWalker]

Thanks for all the info. It's a lot to process but every time I read it it's getting clearer . As for the real flight - today weather got worse but as soon as it's get better I'll try to do some more realistic tests. But for now I think I do have another log that might contain some information. I have added FLY085.DAT file to the OneDrive folder. It should contain the log from that sunflower field I mentioned earlier. I took 180 degree turn there, while flying forward (maybe a bit similar to the turn from the incident flight). I'll upload more logs from real flights as soon as possible.

PS: As for the distance from the incident flight location - it happened while I was on holiday, quite far from home. The best test results would probably require flying in similar conditions in the same location. Unfortunately I won't be able to visiting that place anytime soon. But probably this fall (first half of september) I should be visiting place a lot closer to that location. And even if not, I'll sure be able to fly the drone in place at least 230 km to the north.

The test in FLY085 couldn't be any better.

 

[BudWalker]

@ankalagon would it be possible to get a few of the .DATs that you already have? I'd like to know if your Mavic has always had this problem, did it occur gradually, or was it an abrupt change. If your Mavic has always had this problem I would expect there to have been a previous incident. It helps that you didn't do a calibration prior to the incident.

I recently looked at a P4 incident where it seems there was an abrupt change/failure between flights. That pilot did the same experiments that you did.
Fly Away

Could you retrieve a couple of the the oldest .DATs that you have? And, maybe a couple midway between the oldest and FLY080? I know you're an experienced .DAT retriever by now, but, please make sure the .DATs are actual flights, not the result of turning on the battery for some reason.

 

[ankalagon]

I have uploaded newest FLY088.DAT file from yesterday. I have tried to fly faster with tighter turns, so maybe it will contain some more interesting data.

As for the older flight logs I've uploaded some. I don't understand the content of those, so I've picked them by their modification date and size (ignoring the small ones). The oldest FLY047.DAT was probably made by previous owner of the Mavic (I bought it used from internet auction service around 20 june, so everything created before should be flights made by him).

 

[BudWalker]

Thanks for those older logs. Your Mavic never had a compass calibration issue. It's just that one flight, FLY080, where there was some problem. If there was a compass calibration issue it would have been evident in FLY061, 68, 69, 73, 74. There hasn't been a compass calibration since before FLY047. BTW, FLY047 wasn't an actual flight.

This complicates things. What you'd like to see is a clear indication of a problem in more than one flight, make a change (i.e, the compass calibration), then no problem for more than one flight. Instead, we've got a problem in just that one flight.

Not much else I can add. Quite the mystery. I'm still thinking about it....

 

[ankalagon]

Thanks for all the info and analysis. Although we weren't able to nail the problem, I feel a lot more peaceful knowing that it's probably not factory defect or some kind of hardware failure that caused the incident. Because I bought a used Mavic, I was wondering why previous owner sold it and is there something wrong with it. Now I know that it was random incident. Maybe something more related to the location of the flight. Some magnetic field or electromagnetic interference. As I remember that day there where storms in whole country. Although at my location there was nice sunny day with moderate wind, maybe nearby storms could interfere with the compass. It probably would be best to try flying at the same spot. Unfortunately I won't be near anytime soon.

What I've learned from the incident is that this drones are really complex machines and there is always chance of something going wrong. Too bad that DJI didn't gave the option to manually switch Mavic into ATTI mode because I would love to practice manual flying. When the incident happened I was totally surprised and panicked . It would be good to be able to prepare for such events.

Again thank you for your help.

PS: If you are interested I've uploaded video from the incident flight. Although I doubt if it will cast any more light to the case. No matter how many times I watch it I don't see anything that could cause compass interference.

 

[ThisIsCusel]

Thought I would add to this thread. After many flights without this particular issue, i now have experienced it as well.


DJI Flight Log Viewer - PhantomHelp.com

Flight was in my backyard, outside of losing connection due to crappy DJI remote to phone cable, I've not had any issues flying here prior to this.

Actually, earlier in the day, app lost connection and wouldn't reconnect even after restarting the app (it wasn't frozen, just not connected to mavic) I landed manually, restarted the mavic and it seemed fine after that until later in the day and the afore mentioned atti mode switch

 

[monkeyman]

Hello Gents.

During literally my first flight with the Mavic today I had this same thing happen!

I don't quite know how to post logs or any of that yet as I'm a newbie. If anyone wants me to upload something point me in the direction if it's for the greater good.