Camera system resets mid-flight...?

[Logger]

Thanks for coming back to post in your thread. This is the only way we have a chance to find out what's causing it, everyone posting situations in which it had or hadn't occurred. I was flying in heavy winds today also, so that's not it.

Yes this could help. Also posting your configuration including firmware version, Control app and SD card brand/size may be of use.

I'm not sure if it was discussed before, but another theory I have is that potentially it could be direct sunlight with the drone staying in one place for longer. It's something that was typical for the two flights during which this happened.

I will also discount the sunlight theory as none of my occurrences have been in this situation.

Another thing to consider is: I haven't had any gimbal resets for months of flying it. These two threads dealing with the issue are not very old. Could it be a firmware issue? A memory leak or something similar - in which case ANYTHING could be causing it, and it wouldn't really matter anyway because then it's just the issue with the code in the firmware.

Exactly. There were a few reports of gimbal resets back in December, probably on 0100 and they have started to come back in vogue on 0600 and 0700.

One thing I have noticed is the visible gimbal bumps are not always at the very start of the cache video of the occurrence. Sometimes a couple of seconds in. So the event & therefore one would assume the cause, is actually in process a few seconds before this, at very least when the cache video restarts.

The start and end of this video were dictated by the camera system and not me. ie it reset once at the start and then quit at the end too resulting in a 27 sec recording. Why did it reset at the end?

Was looking at DAT of a reset I had a couple of days ago and the bump was coincident with the aircraft itself pitching down then up and not just the gimbal. Normally the gimbal hides the aircraft movements. But during a reset there is a period when it cannot do this as it has not regained alignment and it seemingly moves with the aircraft. During the yaw left you can see the camera filming the gimbal itself as it had yet to align itself and regain its travel limits. This one also self dictated the recording length being similar at 23 seconds.

BTW I use Samsung 64GB Pro+ class 3 SD card.

 

[Member]

Exactly. There were a few reports of gimbal resets back in December, probably on 0100 and they have started to come back in vogue on 0600 and 0700.

I really like your approach to debugging and the vast info you are providing here. It's much appreciated. I wish I had kept notes about which firmware, SD card etc I was using.

One thing I have noticed is the visible gimbal bumps are not always at the very start of the cache video of the occurrence.

Did you mean the end of the video? For me it's always at the end of the cache (and it doesn't show up on the 4K recording, except maybe the end of the exposure/focus auto-recalibration):

* Flight 1, cached vid 1: It happens 08:49 into the video. Hard & quick gimbal jig, leveled out in 4 seconds, drone flies well for another 16 seconds, then exposure calibration, then re-focus, cached recording stops. I hit record again. 4K recording doesn't show any of this. Recording to the card stops before the jig. This is why the SD card or the filesytem / using a formatted card helping is still a suspect to me. It stops writing the file first, then everything else happens that is captured from the broadcast.

* Flight 1, cached vid 2: Starts with the re-focus from the previous one showing. I fly for 6 more minutes. No gimbal reset but the video stops anyway, obviously I had to hit record again.

* Flight 1, cached vid 3: Nohing special and then gimbal snaps at 01:51 in. Much harder reset than before - not just a quick jig but a more aggressive jig, then it looks up at the props and stays looking at the sky for a second or two, levels out for another few seconds, at 02:12 it does the usual exposure/iso/whitebalance thing, then re-focus, then the video ends.

* Flight 2 with the new drone: I'm 09:04 into the cached video. It does the usual jig, the horizontal calibration takes a few seconds but doesn't level it out properly, then big exposure as before, then re-focus starts, video ends, focus is still a bit off in the next cached video but I don't even bother with it, I just bring back the aircraft as quickly as possible.

Unfortunately I'm not 100% sure which SD card I was using at the time. I think I was using the same 64G Sandisk SD card (not the cat3 extreme-hyper-super-duper-made-for-4K one, just the 'regular' one, the type DJI ships the drone with, except not the 16 gig version).

Much of what is happening after the jig I didn't see at the time I was flying, I just remember losing the live picture and control and not being sure if the image on the screen is stuck or whatever. During the first flight it did say signal lost also, I'm not 100% sure about it. Traumatic event, got erased from the brain LOL. When you lose control of your drone and have no idea what's happening, it feels like minutes have passed with every fqn second. People say this a lot but I specifically remember thinking "the drone is lost, what's my next step to get it back" just before the picture came back, so I think I lost control for a good 5-10 seconds at least. After reading your report, I'm curious to see what happens in such an event to the bird in the air. It's scary to think that it would do anything other than hover in one place...

 

[Logger]

Interesting @Member. Seems your gimbal event cache recording are getting packaged differently to mine. Are you per chance on IOS and me on Android?

In my events I get three cached segments. Before event and After event which appear normal with a seperate 25-30 segment encompassing the reset.
In yours it seems the cache does not get interrupted at the event start and gets chopped at the end only. Could be an OS issue or even a portable device issue regarding cache behaviour.

Your comments wrt to SD cards getting written to just twigged something. First event above definitely occurred when my 2.7 recording hit the 3.80GB limit and started a new file. Second occurrence occurred precisely when I manually stopped recording and toggled to Photo to take three still images.

So in both instance a major change occurred wrt to what was going on with the SD card. Perhaps writing a new file and something went haywire.

 

[Member]

Interesting @Member. Seems your gimbal event cache recording are getting packaged differently to mine. Are you per chance on IOS and me on Android?

Nope. I'm using a Samsung Galaxy Tab S2 - the only tablet I know of that isn't Apple and fits in the Mavic controller's holder (I bought it specifically for the Mavic). Screw flying with a phone sized live picture. Even with the tablet and the beautiful screen it has, it's still hard to see cables and wires. But it looks great and I can see it very well even when it's sunny.

In my events I get three cached segments. Before event and After event which appear normal with a seperate 25-30 segment encompassing the reset.

Interesting.

In yours it seems the cache does not get interrupted at the event start and gets chopped at the end only.

It chops the recording fully, but the video cache is only interrupted by the end of the re-focusing stage.

Could be an OS issue or even a portable device issue regarding cache behaviour.

Or DJI Go version maybe. I'm on DJI Go 4 v4.0.2.

So in both instance a major change occurred wrt to what was going on with the SD card. Perhaps writing a new file and something went haywire.

Yep, pretty much.

1st flight, 1st video: 4.1 GB (ends just before the jig in the cached video shows)
1st flight, 2nd video: 4.1 GB (starts just after the re-focus, ends just before the jig)
2nd flight, 1st video: 4.1 GB (ends just before the jig)

We might just be getting somewhere, eh?

Same folder (same day), two flights with two batteries totalling 5 videos with the newly bought Sandisk 32 GB ultra-super-duper-extra-4k-hispeed microSD card:
- 4.1 GB
- 3.3 GB
(switching the battery)
- 4.1 GB
- 4.1 GB
- 1.0 GB
(second battery finished)

Overall: no problems since the card was replaced.

Short of fixing the firmware, I think one of these two things may be a workaround:
1) use a for-4K, class-3, <=32 GB card
2) format the card every time before you go out to fly, don't just move the files.

I'm positive that the culprit is less likely to be the card class, but rather the file system, the index files left there or something similar. *EDIT* @HSK said "I format my card each time I insert it." - if that is true, including when he had the reset, then I guess we're back to square one?

And finally, one more debugging question for you: what OS are you using to copy/move the files? I'm on Ubuntu.

 

[_milo_]

Interesting @Member. Second occurrence occurred precisely when I manually stopped recording and toggled to Photo to take three still images.

So in both instance a major change occurred wrt to what was going on with the SD card. Perhaps writing a new file and something went haywire.

That's interesting, I've had 3 gimbal resets on across different firmwares, and I believe my first one happened exactly like this, I stopped video, and triggered photo, probably was a bracketed 3 shot photo.

My last one was on .600 and at the 4gb limit,

I was using Sandisk Extreme Pro U3 64gb

 

[Logger]

Nope. I'm using a Samsung Galaxy Tab S2 -
<snip>
Or DJI Go version maybe. I'm on DJI Go 4 v4.0.2.

Yeah it could well be the DJI Go 4 version that is varying the cache behaviour.

Yep, pretty much.

1st flight, 1st video: 4.1 GB (ends just before the jig in the cached video shows)
1st flight, 2nd video: 4.1 GB (starts just after the re-focus, ends just before the jig)
2nd flight, 1st video: 4.1 GB (ends just before the jig)

We might just be getting somewhere, eh?

Might be. It is looking suspiciously like it has to do with the firmwares file writing routine.

Short of fixing the firmware, I think one of these two things may be a workaround:
1) use a for-4K, class-3 card
2) format the card every time before you go out to fly, don't just move the files.

I'm positive that the culprit is less likely to be the card class, but rather the file system, the index files left there or something similar. *EDIT* @HSK said "I format my card each time I insert it." - if that is true, including when he had the reset, then I guess we're back to square one?

I am not sure about the format method and its involvement. I only ever format my card with the Mavic itself. Probably about every third 1/2 dozen flight. Sometimes I stick the card in my PC and view stuff and delete one or two unwanted recordings and stick it back in the Mavic like this.

And finally, one more debugging question for you: what OS are you using to copy/move the files? I'm on Ubuntu.

PC Windows 10 and I never format SD card with it.

 

[Logger]

That's interesting, I've had 3 gimbal resets on across different firmwares, and I believe my first one happened exactly like this, I stopped video, and triggered photo, probably was a bracketed 3 shot photo.

My last one was on .600 and at the 4gb limit,

I was using Sandisk Extreme Pro U3 64gb

Doubly interesting. I have had four gimbal resets in total . Two in the past week mentioned above and another two back in early April. Just went and reviewed the earlier ones which were on 0600. Again both occurred at file creation events. One was the 4gb limit and the other when I manually stopped recording to hover and took ake some stills. So I have a 100% strike rate for a new file being created right at the point of Gimbal reset.

Reported reset events have occurred with files less than the 4gb limit size, when changing from video to still image mode. So the file limit is probably not the the cause. It could mean it is actually the process of initiating the creation of the subsequent file that is the trigger. So 4gb limit causes new file creation, just as stopping video and taking a still triggers a new file creation? What is going on in the image system as this is occurring? Is something in the image system becoming overloaded?

Have all the reported inflight gimbal reset occurred when either, recording video, or switching to and taking stills after a period of recording?

Put another way - has anyone had an inflight gimbal reset while only using the camera for FPV and not saving anything to the onboard SD card.

 

[Member]

When a file is closed and/or a new file is written, there's some other stuff done like maybe thumbnails generated, the index files updated and whatnot. The bug could be there. So it would be triggered when the file limit is reached as well as when you simply stop recording or even switch to the camera.

Do any of you remember any reports of this happening with a card that wasn't 32 GB or less in size? All the reports I remember (and I definitely don't remember them all) talked about 64 GB or 128 GB.

 

[Logger]

When a file is closed and/or a new file is written, there's some other stuff done like maybe thumbnails generated, the index files updated and whatnot. The bug could be there. So it would be triggered when the file limit is reached as well as when you simply stop recording or even switch to the camera.

Yep.
For me, when it has occurred after a switch to camera I have managed to capture several images at least. Another thing that only started working properly in a recent firmware was Auto Syns HD photo, so sending file from card to mobile device.

Do any of you remember any reports of this happening with a card that wasn't 32 GB or less in size? All the reports I remember (and I definitely don't remember them all) talked about 64 GB or 128 GB.

Only ever used a 64GB card.

 

[Strafe1]

I'm probably out contributing to this conversation but I'll mention that I have never had a gimbal reset mid flight. I use a Samsung Pro + 64G card and remove the card from the Mavic to download it onto my Mac, from which I then erase the card before ejecting it. I update FW as it's released and I use an iOS device. Hope this helps someone!

 

[alexcannas]

I think it has been well documented that the weakest link in the gimbal is the ribbon cable. I've said it before and I'll say it again the firmware cannot be the cause. It may alter the behaviour once it regains signal, but that's really something that should not be happening in any case.

Save yourself the trouble and either send the Mavic back for repair or replace the ribbon cable. Note that the are two different cables, the ribbon only controls the gimbal. The camera is connected with a more robust wire cable.

@HSK The test with the fan blowing onto the Mavic is ingenious and proves the point.

 

[HSK]

Agreed...As much as I'd love it to be a software bug that could be solved with a firmware update, It's gotta be hardware issues.

I think DJI just simply wanted to get these units out the door to meet demand, and just relaxed the QC/QA process...with a "just deal with the returns" mentally.

It may have improved now, I have a new Mavic which I've subjected to exactly the same tests and conditions as the last, and it is absolutely fine right now.

 

[EsiMavic]

Also my c1 and c2 button has a slow reponse after i experience the gimbal reset

 

[EsiMavic]

But why is it some didnt experience it when they downgrade to .400??

 

[Logger]

... I've said it before and I'll say it again the firmware cannot be the cause. It may alter the behaviour once it regains signal, but that's really something that should not be happening in any case.
Save yourself the trouble and either send the Mavic back for repair or replace the ribbon cable.....

How does my ribbon cable know to only only ever fault and reset my gimbal at the exact moment my recording reaches 3.8gb? It also knows to reset my gimbal only at the exact moment my firmware writes a new image file after having switching from recording mp4 files. Intriguing given no video or imagery passes along said cable.

There is a common theme developing as to when these resets occur. The common timing of these resets suggest the trigger is how a (firmware) process interacts with the hardware (Gimbal - imaging system - processors - SD card media and cables connecting same).
You say the firmware cannot be the cause. I say the gimbal control ribbon cable us unlikely to be the cause. If it were, logic dictates the events would be randomly dispersed as the cable flexes and moves and not occurring concurrently with a specific file process.

For me the resets have been infrequent. 4 in 605 flights. They are at worst a bit of a nuisance. I dont want to go to the trouble of sending the Mavic back for repair or to do a self replacement the the ribbon cable when I cannot see how it can be the cause.

 

[HSK]

my experience - camera gimbal would reset at any given time. you couldn't determine when it would happen. Anything from a few mins to over ten mins. and then sometimes it would go all the way to 30 mins with no issue. I'm trying to think of the interactions going there. certainly, only with a few moments of recording, nothing would be going on like file splitting and thumbnail writing....apart from just recording itself.

 

[alexcannas]

But why is it some didnt experience it when they downgrade to .400??

That may be circumstantial. As @HSK said in his case it didn't make a difference.

The problem is that the cable is not broken but does not make good contact. That means the issue may appear due a number of factors, like air pressure, heat or vibrations.

I would also expect that if this was caused by bad firmware, there would be a lot more people (if not all) experiencing it.

 

[Logger]

my experience - camera gimbal would reset at any given time..

Then in your particular circumstance it was seemingly random and is indeed hard to attribute to a specific event. A new unit fixed things for you. We will never know what component was faulty though. Could have been one of the two cables, could have been the Gimbal control board or one of a multitude of things deeper within the system.

For the rest of us, having the gimbal resets occurring in coherent situations, suggests it is not a random momentary loose cable.

I would also expect that if this was caused by bad firmware, there would be a lot more people (if not all) experiencing it.

The flaw in this argument is that not everyone (and in fact very few people) will have their firmware configured in the same manner.

Are your settings exactly the same as mine? Here is just one page of my multitude of configurable settings.

Who is to know what esoteric setting, when combined with a particular flight event like file swap and high CPU or memory load might trigger an event. There are probably plenty of people who do not fly around filming continuously either. When have you seen it occur on your Mavic exactly @alexcannas ?

As I recall Autosync HD photo only started working properly in the last couple of months with a firmware update. The setting was there but it never actually worked properly. This file related event could be a trigger. AFC is also relative recent. That means AF will occur at the start of a new recording or some such and when taking stills. Another possible trigger. Lock gimble when capture.... Hmmm what if a set of circumstances caused it to not unlock after capturing a still and had it move with the aircraft pitch, as shown in my video above. So many possibilities.

Anyhow I do not think it is necessarily caused by the firmware. Just that I am not precluding it.

 

[EsiMavic]

I guess it occur in .550. But i dont want to downgrade t0.400.but when i updated to .700 it is the first time that i reach the 9 mins video recording and i guess it stops recording because of the 4gb split. I still experience the random stops though. Ill be happy if my only issue is the video split, to me recording 9mins plus is enough. Who wants to see a 9 min video of the same subject same location and same angle shot??

 

[Logger]

I guess it occur in .550. But i dont want to downgrade t0.400.but when i updated to .700 it is the first time that i reach the 9 mins video recording and i guess it stops recording because of the 4gb split. I still experience the random stops though. Ill be happy if my only issue is the video split, to me recording 9mins plus is enough. Who wants to see a 9 min video of the same subject same location and same angle shot??

Sorry @EsiMavic I inadvertently quoted you when I meant o quote @alexcannas

But all good. Your 9 min incidents also tie in perfectly with the file split. So unlikley to be a ribbon cable. Next time you see a random stop see if you can recall what was going on at the time. Is it mid recording or perhpas have you just switched from recording to stills.

I agree no one is going to watch a 9 min video. Although an exception is when you are flying a route to check things in detail like a farm. I watch it once then archive or delete it.

I am happy to stay with .700 too.