Phantom 2 Catastrophic Failure after Firmware 3.06

[Lozboz]

I seem to recall a similar vote of confidence in the latest firmware from Kelly last Monday. Only to be sadly followed by his account of it diving into the ground that day.

I really hope that nintendrone64 doesn't have the same experience, we are losing too many. In the main they too appear to be experienced pilots.

IMHO no amount of reading the manual can help you when the 3.06 firmware decides, for whatever reason, that it's coming down. It is a horrible feeling.

I really would recommend going back to 3.04 whilst it gets sorted.

 

[thongbong]

Fun fact: I've had a Vision and Vision+, my Vision since December. Not only did I fly it in a mall for the first 3 months as a job, I did not fly it with GPS mode on. I always upgrade to the most recent firmware and I have never experienced a single problem that so many folks on here claim to have. It's not DJI's fault that you're not reading the manual.
I take my craft over water when I want to, and I never have to worry about it just "dropping from the sky" because that doesn't "just happen". Sorry, mate.

\: *world's smallest violin plays"

LOL, you just jinxed yourself.

 

[ianwood]

Kelly's crash was very different from xrover's. At least in description. We do not know what caused either of these two incidents. We have some suspicions that need to be verified or dispelled. That they both happened on 3.06 could lead to a root cause or it could simply be coincidental. FWIW, the battery issues that I am aware of precede 3.06.

 

[wkf94025]

The video of my v3.06 FW crash is below.

Recap of earlier comments: Following the upgrade to v3.06 a week prior to the crash, I did full calibrations in Assistant, including IMU, and compass calibration. I had flown v3.06 a dozen times prior to the crash, with no issues whatsoever. The site where the crash occurred I have flown and filmed at least 50+ times since my first Phantom 1.0 almost a year ago. This P2 and H3-3D and accessories had been flying just fine several times a week on 3.04 since early April, including dozens of flights on the site of the crash. The setup was new in box in early April, and had never been crashed until this fateful day. On the day of this crash, I had lock on 7 sats at takeoff, then home point set, and 10 sats once stabilized in the air. The only change between the fatal outing August 11th and ~50+ prior flights on the exact same hardware is 3.06 versus 3.04. And the departure behavior is so indicative of a flight controller than suddenly decides it should be somewhere else. My current theory is that GPS / NAZA are badly confused all of a sudden. Also, FWIW, I have had GPS go out on multiple occasions on previous P2's running v3.04. The behavior in those cases was much more benign than in this case: the flight mode on OSD suddenly switches from GPS to ATTI, and the # sats shows zero. However, in the case of v3.04 losing GPS, there is a graceful transition from GPS to ATTI, not a sudden departure to an unknown coordinate system. Also, having had battery premature auto-landing experiences, power subsystem failures, and bad ESC's, I don't see indications of any of those symptoms in this crash.

HAL lives on in v3.06 in my opinion.

Kelly

[youtube]https://www.youtube.com/watch?v=fkmvUfC0Ov0[/youtube]

 

[ianwood]

Interesting. Pretty stable in the beginning but were you constantly making small adjustments before the flyaway? It seems the P2 never stopped moving. Wondering if that was you making small adjustments or it was moving slightly on it's own.

What I do see that is interesting is the gimbal loses the horizon in perfect sync with the flyaway and it leans to the right by the same amount as the Phantom. I think that is usually characteristic of an IMU issue. What I suspect might have happened is the IMU for whatever reason thought that it is was tilted to the left and tried to compensate by instructing the P2 to lean to the right. It instructed the gimbal to do the same so it could also stay level. This causes it to fly off in that direction slowly towards the ground with the gimbal equally tilted.

I think you need to contact DJI and show them this. The root cause will be hard to determine but it is clearly a classic flyaway. We have seen IMU related flyaways like this before 3.06. They're just not that common anymore thankfully.

 

[hrosee]

I think it's possible to get a bad firmware upload and not know it.

After I went to 3.06 I calibrated the transmitter in the RC assistant and then everything in the Phantom assistant. When I tried to reset the course lock or home lock using the left switch on the ground or in the air it would not work except about every 10th time. I reloaded the firmware and it works flawlessly now.

On each flight now I take it up fly it out a ways, flip it into CL and may sure it goes the direction it is supposed to go. Then I do the same for HL. There has been a few times that HL takes it off somewhere else. This means it didn't set the home position correctly. When this happens I bring it back, land it, set the home position manually, then do another test. It has always worked the second time. It only takes about 20 seconds to test this and is well worth it. If anything goes wrong and the home point is not set correctly you will have a flyaway to wherever it thinks the home point is.

I suggest everyone do this quick test no matter what version you are on.

Also I have had a couple of times where the battery has gone real low and it starts a decent on its own. This happened to me today when I let the battery get to 10 percent. I gave it more throttle and it kept coming down. I then flipped the right switch between GPS and ATTI and regained control. All was well and I regained control and landed with only 9 percent battery left.


There are many ways to experience a flyaway that is not really a flyaway. Today I was playing with ATTI and was also in course lock. I had passed the point and now everything on CL was backwards which is normal. I can tell you if you get in that position with CL and ATTI and fly over you things get real weird but its doing what you are telling it to do.

To recover it seemed logical to me to flip into home loch. But to my surprise HL does dot work when you are also in ATTI. I flipped the switch back to GPS and then HL worked just fine. I got it back to the point where I could see it and reset CL in the air for the new direction and all was well again.

BOTTOM LINE, if you cannot reset CL and HL on the ground you need to reload the firmware or figure out what else is wrong. SECOND, always go out a short way and make sure that CL and HL act right. If not bring her back and reset it manually on the ground. This extra minute of battery use sure beats a flyaway. Finally, if you feel you are having throttle problems try and flip the right switch between gps and atti. You just might regain control. Mit worked for me.

I hope all this makes sense. If not ask away.

 

[wkf94025]

Ian,

Good eye. see my comments inline in red.

Kelly

Interesting. Pretty stable in the beginning but were you constantly making small adjustments before the flyaway? Yes I was making constant adjustments to frame the shots and do circular pans around the excavators.It seems the P2 never stopped moving. That was active input by me. Wondering if that was you making small adjustments or it was moving slightly on it's own.

What I do see that is interesting is the gimbal loses the horizon in perfect sync with the flyaway yes, I noticed gymbal twitching simultaneous with the start of the flyaway and it leans to the right by the same amount as the Phantom. Good eye! I think that is usually characteristic of an IMU issue. What I suspect might have happened is the IMU for whatever reason thought that it is was tilted to the left and tried to compensate by instructing the P2 to lean to the right. It instructed the gimbal to do the same so it could also stay level. This causes it to fly off in that direction slowly towards the ground with the gimbal equally tilted. I think you might be on to something here...

I think you need to contact DJI and show them this. The root cause will be hard to determine but it is clearly a classic flyaway. We have seen IMU related flyaways like this before 3.06. They're just not that common anymore thankfully.

I would be DELIGHTED for them to diagnose this as unrelated to v3.06, so long as they have some idea how to root-cause and rectify this. My other P2 has only 2 flights under its belt, and I'm not risking it until they give me a sense for why the other one bit the dust like this.

 

[Fyod]

All of these cases are very similar. Yours is VERY similar to the guy flying in a city above the granite walkway, especially in that both of you seem to be encountering little, if any wind.
It is almost always the same - calm flying, all of a sudden very hard tilt at an aggressive angle, crash.
I doubt so many people would have some kind of magnetic issue or forgot to recalibrate the compass.
Personally, I haven't recalibrated the compass for about 2 months, with multiple flights during that time with no problems.
I say there is a major flaw in 3.06 and all of you guys should contact DJI so they see that the problem is redundant.

 

[ianwood]

All of these cases are very similar. Yours is VERY similar to the guy flying in a city above the granite walkway, especially in that both of you seem to be encountering little, if any wind.
It is almost always the same - calm flying, all of a sudden very hard tilt at an aggressive angle, crash.
I doubt so many people would have some kind of magnetic issue or forgot to recalibrate the compass.
Personally, I haven't recalibrated the compass for about 2 months, with multiple flights during that time with no problems.
I say there is a major flaw in 3.06 and all of you guys should contact DJI so they see that the problem is redundant.

The granite one had some TBE in it if I remember correctly. It was most likely caused by compass discrepancy at initialization. (I can only guess though.) I don't see any TBE in Kelly's video so I am not sure what role, if any, the compass has in this flyaway. But they are related in that the TBE ones do have the same loss of horizon in the gimbal. The gimbal going sideways along with the Phantom tells me the Phantom can't balance, like it has an inner ear infection! That would be consistent with why it bolts in one direction.

This could be figured out but we need definitive answers from DJI on how all these sensors interact, how they initialize, what happens in these situations.

 

[neoben]

I'm no expert in these things.... but doesn't the fact that the camera, which is maintaining level throughout the whole flight, suddenly decides to roll with the Phantom during the crash indicate 2 things:

1) This is not operator error. There's nothing I can think of that an operator would do that could cause that to happen
2) Heli's level and camera level are controlled by the IMU, which was obviously working up to that point... (I assume just 1 IMU for both Phantom and Gimbal, right? Like on a P2V+? Or are there 2 on this setup? That would change things).

So, the conclusion is that it was an IMU failure of some degree... Could be the physical IMU connection became weak over time and at that point finally gave up and either popped out (if it's a connector) or came unsoldered (if not).

A compas/gps/user error would not cause the camera to roll like that.

 

[ToddSmi]

The video of my v3.06 FW crash is below.

Kelly, really sorry to hear of your crash. This does appear to be the classic type "flyaway", unlike many we hear of. I would definitely share this footage with DJI and hopefully they can learn something from it as well as step up and take care of you.

On a side note, is the video slowed down a touch? The sound/pitch from the motors sounded like normal speed, but the movement of the tractors in the beginning looked possibly on the slow side so I was just curious if this was 100% speed or not, for those that may want to look for clues with the original audio. If that was normal playback speed, then doesn't really seem like much out of the ordinary with the motor sounds either, further pointing to classic "flyaway". Also, did you happen to notice if the telemetry was indicating GPS or ATTI mode or flickering at time control was lost?

 

[isopro]

The video of my v3.06 FW crash is below.

Recap of earlier comments: Following the upgrade to v3.06 a week prior to the crash, I did full calibrations in Assistant, including IMU, and compass calibration. I had flown v3.06 a dozen times prior to the crash, with no issues whatsoever. The site where the crash occurred I have flown and filmed at least 50+ times since my first Phantom 1.0 almost a year ago. This P2 and H3-3D and accessories had been flying just fine several times a week on 3.04 since early April, including dozens of flights on the site of the crash. The setup was new in box in early April, and had never been crashed until this fateful day. On the day of this crash, I had lock on 7 sats at takeoff, then home point set, and 10 sats once stabilized in the air. The only change between the fatal outing August 11th and ~50+ prior flights on the exact same hardware is 3.06 versus 3.04. And the departure behavior is so indicative of a flight controller than suddenly decides it should be somewhere else. My current theory is that GPS / NAZA are badly confused all of a sudden. Also, FWIW, I have had GPS go out on multiple occasions on previous P2's running v3.04. The behavior in those cases was much more benign than in this case: the flight mode on OSD suddenly switches from GPS to ATTI, and the # sats shows zero. However, in the case of v3.04 losing GPS, there is a graceful transition from GPS to ATTI, not a sudden departure to an unknown coordinate system. Also, having had battery premature auto-landing experiences, power subsystem failures, and bad ESC's, I don't see indications of any of those symptoms in this crash.

HAL lives on in v3.06 in my opinion.

Kelly

[youtube]https://www.youtube.com/watch?v=fkmvUfC0Ov0[/youtube]

Could this be a crash caused by an ESC failure, if not why not ?

 

[wkf94025]

Toddsmi,

The video should be regular speed. GoPro captured at 1080p/60fps, rendered at 30fps, but other than that no intentional speed change. i didn't notice GPS/ATTI flicker, but at time of fly away was looking at other tractors and getting ready for next shot, so cant say with 100% confidence.

Isopro,

i suppose it is *possible* its an ESC failure, but I think Ian's theory re IMU is much more likely. The ESC failure I had in April was much more dramatic in its impact. Full on loops/rolls as one leg of the chair was cut off.

Kelly

 

[vudu]

Having read this thread from the beginning I'm coming to believe that these issues don't have anything to do with a specific version of the firmware. I've had a similar flying experience that resulted in near-crash but fortunately I was able to catch-land the bird safely in the end.

[youtube]http://www.youtube.com/watch?v=WBwPxRqVhVU[/youtube]

Prior to the footage I was filming traffic for about 6 minutes and P2 was pretty stable (apart from compensation for the wind which was pretty strong that day). Then some heavy TBE started to occur and I decided to stop filming and land. The symptoms were similar to those on the video above: uncontrolled fast sideways movement with gimbal not compensating the tilt. The only difference is that I was at a higher altitude and P2 didn't crash into anything when it went haywire. I landed it in GPS-mode. I didn't use HL because the spot was pretty tight and there wouldn't be much use to it.

The thing is I was flying with 3.04 at the moment. Yesterday I updated to 3.06, passed all the calibration procedures and test-flew 2 batteries. No troubling symptoms occur, the overall stability seemed to be better, no TBE or anything of the sort. I hope it remains this way.

 

[Xrover]

Kelly,
Sorry about your crash

Your video really shows the classic Fly Away. There are definitely more then one type of Phantom failures blanketed under the so called Fly Away. Yours shows the failure where the Phantom will just make a break for it as if it was busting out of jail.

Mine Just lost power and went straight down.

I think we all should be breaking up the different types of Fly Aways so we can break the issues down and hopefully get a future firmware correction.

So what I have so far is:

1. Jail Break
2. Pelican Dive. ( copyright Xrover lol)
3. Early Autoland

 

[wkf94025]

Jail Break and Pelican Dive. Like it. Maybe DJI will sponsor a name-that-crash contest on the front page of their web site! Winner gets a free prop signed by the firmware engineering team.

Kelly

 

[Xrover]

Jail Break and Pelican Dive. Like it. Maybe DJI will sponsor a name-that-crash contest on the front page of their web site! Winner gets a free prop signed by the firmware engineering team.

Kelly

That's funniest thing i've heard all month!! :lol:

 

[droneranger]

Jail Break and Pelican Dive. Like it. Maybe DJI will sponsor a name-that-crash contest on the front page of their web site! Winner gets a free prop signed by the firmware engineering team.

Kelly

That's funniest thing i've heard all month!! :lol:

+1

 

[tferrari]

I'm flying 3.06 with J-hooking and TBE solved. This has been great for me in Northern Canada with high magnetic declination.

Adjusting the gains higher than what the stock settings were (for basic gains) has really improved overall flight stability.

 

[Xrover]

3.06 Definitely solved the J Hook and TBE issues..