Spooky, first time i've almost lost it, what happened?

[CRCs Reality]

So, after almost 80 flights with zero issues at all I had a scary one today.

Got to the courthouse early, and figured I'd launch from the roof of the parking garage (yeah, I know) and get a nice shot of town above the courthouse.
I was very careful to check for any compass errors (again, parking garage), but all looked good so I went up to just below 400' and started to move into position for a panorama shot.

At that point, the status went from "P-GPS" to "ATTI" and yellow, and the P3 started drifting with the slight breeze. I lost most control, but not quite all.
Thankfully, I had seen many posts where people panicked and hit RTH in this situation, causing a flyaway (can't RTH with no GPS, right?) so I switched to ATTI mode and was able to slowly bring it back over me and down for a landing.

Here's the problem(s):
1 - when coming back in ATTI mode, the controls were VERY sluggish, I had to peg the controls to get much movement at all, and wind wasn't a factor. It was really weird, I've flown in ATTI before to get used to it, and something was definitely wrong.

2 - Looking at the HealthyDrones logs, I expected to see GPS problems, but there were only a few (10-15 instead of 15+) and only close to where I landed.. Through most of the loop where I was forced into ATTI and had very little control, I had solid GPS and Compass readings. ( Healthy Drones Logs of the flight)

It's shaken my confidence a little, I do have the .dat file from the drone, but really don't know what to do with it to try and unravel this mystery..

Any input from those with more flight time than me as to WTF happened?

 

[CRCs Reality]

Oh, latest firmware on the bird, batteries, and controller if that matters.
Also, I was using Litchi (latest version) for the flight, but doubt that was a factor.

I have to point out, if it weren't for the many different posts on here urging folks to get familiar with ATTI mode in case they need it, I might have panicked into a RTH.. Which would have made this a much less happy thread about my flyaway instead of just me wondering what went wrong (but happy I got it back without a scratch).

 

[msinger]

Please upload your TXT flight log here and post a link back here.

 

[CRCs Reality]

Please upload your TXT flight log here and post a link back here.

Phantom Log Viewer - PhantomHelp.com

Appreciate the help!

 

[CRCs Reality]

Interesting logs... I can see it bouncing between P-GPS and ATTI (on it's own, not from me) with plenty of satellite signals..

 

[Digdat0]

Interesting logs... I can see it bouncing between P-GPS and ATTI (on it's own, not from me) with plenty of satellite signals..

I wonder if the latest Dji go update gps fix is intended for situations like this.

 

[Meta4]

Got to the courthouse early, and figured I'd launch from the roof of the parking garage (yeah, I know)
At that point, the status went from "P-GPS" to "ATTI" and yellow, and the P3 started drifting with the slight breeze. I lost most control, but not quite all.

What exactly happened up there on the parking garage?
Did you recalibrate the compass there?

 

[shockwave199]

Got to the courthouse early, and figured I'd launch from the roof of the parking garage (yeah, I know) and get a nice shot of town above the courthouse.
I was very careful to check for any compass errors (again, parking garage), but all looked good

Yeah you know, lol. My opinion is that your launch point played a major role in the problem, even if records deem it undetermined. There isn't a single good reason to do that, not one. I ran into the same kind of trouble a while back when launching from a mere sidewalk that was surely chock full of rebar. I will never make that mistake again. I knew better and yet somehow I did it. The main thing is to learn from it. Glad it worked out ok.

 

[neven]

I must admit that I'm still confused why a launch from surfaces with metal inside would cause problems later in flight. If you calibrate there I understand. Maybe some problems while take off and landing I also understand. But why would that cause problems in flight hundreds af meters away?


Sent from my iPhone using PhantomPilots

 

[CRCs Reality]

What exactly happened up there on the parking garage?
Did you recalibrate the compass there?

Nope, didn't re-calibrate at all.

And I was hesitant to launch there because of everything I had read about compass interference from structures like that.. After launch I hovered for a moment to make sure everything was OK. It was only once I got some altitude that it got sketchy.
Maybe the fact that I launched from there was the issue, but it makes no sense to me why it would only have been a control problem a few hundred feet, and not 15 feet up

And, IF that's what the problem was, why no compass errors (or any others) in the logs? From what I can see, the logs look like I spent the flight switching between P and ATTI modes manually, which only happened one time in an effort to bring it back down.

Weird..

 

[BudWalker]

I must admit that I'm still confused why a launch from surfaces with metal inside would cause problems later in flight. If you calibrate there I understand. Maybe some problems while take off and landing I also understand. But why would that cause problems in flight hundreds af meters away?


Sent from my iPhone using PhantomPilots

The P3 gets it's heading info from a combination of sensors; mostly the gyros, then the accelerometers. Shortly after batteryOn the Flight Controller initializes the Yaw value based on the compass magnetometers but after that it's info coming from the gyros and accelerometers that determine After this the magnetometers have only a small effect on the value of Yaw. Yaw.

If the P3 launches from a man hole cover that initial Yaw value will be incorrect because of the geomagnetic distortion and it's effect on the magnetometers. When the P3 has clears the effects of the man hole cover the compass/magnetometers become correct, BUT the Yaw is being determined by the gyros which hasn't changed.

Here's an example of where that happened

The P3 was launched from a concrete driveway that contains rebar. magYaw is a diagnostic value calculated from the magnetometers. magYaw and Yaw (the value that the P3 uses for heading) are the same incorrect value until the P3 launches at time 7.4. Then magYaw (blue line) changes to -162, the correct value, BUT Yaw stays at the old, incorrect, value of -27. You can also see the effect on the magnetometers (red line) as the P3 leaves the effects of the rebar. This is the reason that a COMPASS_ERROR_LARGE is declared (denoted by the blue background).

Later, at time 71 the COMPASS_ERROR_LARGE had been replaced with a YAW_ERROR_LARGE

The pilot initiates RTH shown off-blue background but the P3 still has incorrect Yaw info. The P3 then towards what it thinks is home. It was 2 minutes before the Yaw was finally corrected.

 

[neven]

The P3 gets it's heading info from a combination of sensors; mostly the gyros, then the accelerometers. Shortly after batteryOn the Flight Controller initializes the Yaw value based on the compass magnetometers but after that it's info coming from the gyros and accelerometers that determine After this the magnetometers have only a small effect on the value of Yaw. Yaw.

If the P3 launches from a man hole cover that initial Yaw value will be incorrect because of the geomagnetic distortion and it's effect on the magnetometers. When the P3 has clears the effects of the man hole cover the compass/magnetometers become correct, BUT the Yaw is being determined by the gyros which hasn't changed.

Here's an example of where that happened
View attachment 61482
The P3 was launched from a concrete driveway that contains rebar. magYaw is a diagnostic value calculated from the magnetometers. magYaw and Yaw (the value that the P3 uses for heading) are the same incorrect value until the P3 launches at time 7.4. Then magYaw (blue line) changes to -162, the correct value, BUT Yaw stays at the old, incorrect, value of -27. You can also see the effect on the magnetometers (red line) as the P3 leaves the effects of the rebar. This is the reason that a COMPASS_ERROR_LARGE is declared (denoted by the blue background).

Later, at time 71 the COMPASS_ERROR_LARGE had been replaced with a YAW_ERROR_LARGE
View attachment 61484

The pilot initiates RTH shown off-blue background but the P3 still has incorrect Yaw info. The P3 then towards what it thinks is home. It was 2 minutes before the Yaw was finally corrected.

Thank you for this explanation. It makes sense. Although I don't understand why DJI made it work like that because looks like error could be reported to pilot within few seconds after liftoff.


Sent from my iPhone using PhantomPilots

 

[BudWalker]

Thank you for this explanation. It makes sense. Although I don't understand why DJI made it work like that because looks like error could be reported to pilot within few seconds after liftoff.


Sent from my iPhone using PhantomPilots

It was reported.

 

[neven]

It was reported.

In the flight from your example or OP? Because most of Compass errors reported here are midflight ones (mine was 2 minutes into the flight and it was Yaw_large).


Sent from my iPhone using PhantomPilots

 

[BudWalker]

In the flight from your example or OP? Because most of Compass errors reported here are midflight ones (mine was 2 minutes into the flight and it was Yaw_large).


Sent from my iPhone using PhantomPilots

In the flight from the example. DJI makes things very confusing. COMPASS_ERROR_LARGE, YAW_ERROR_LARGE, and SPEED_ERROR_LARGE are values that the field nonGPSCause can have. In flight the DJI Go APP will say "compass error" when the nonGPSCause field has one of these values. But, depending on the Go App version, the flight replay will show these values as something like "GPS lost" (I forgot the exact wording and it has changed between versions). The first time I saw this in flight it wouldn't show anything when I replayed the flight. Six months later, with a newer version of the Go App, on replay it showed as the "GPS Lost".

We're having to speculate what the exact meaning of these values are. I'm pretty confident that COMPASS_ERROR_LARGE means an abrupt change in the magMod value like in this example. I'll speculate that SPEED_ERROR_LARGE means a difference in expected vs. actual velocity or position. I think YAW_ERROR_LARGE means a difference between Yaw and some other measure that the FC has. magYaw was implemented in DatCon to help determine the meaning of YAW_ERROR_LARGE. But, in your flight magYaw didn't help much.

 

[CRCs Reality]

Wow BudWalker, thanks!
I really appreciate the detailed explanation, and that makes sense I suppose.

I can say that from my flight, there were zero compass errors reported to me by the software. And, the only time it went to ATTI mode on screen was after I was already over 300 feet up. Had it given me any indication of a problem early on I'd have brought it back down sooner. Even the log files don't seem to indicate any compass issues at all, which is why I'm confused on the whole situation.
I opened the logs fully expecting compass issues and was blaming myself for taking off from the parking garage when I knew it could be an issue. It was only when I couldn't find any indication of those problems I reached out for help.

A couple of months ago, I took off from a railroad bed that I didn't realize was littered with pelletized iron ore (looked like odd, round gravel). But I was immediately notified within 2 feet of launch that there were compass issues. I landed without incident, and after checking the environment a bit realized my mistake and chalked it up to a learning experience
Because of that, I was actually watching specifically for compass issues on this recent flight, and when I saw none I figured I was OK to go ahead.

Either way, your explanation makes sense (except for the lack of any compass errors in the logs) and I suspect is exactly what happened. I'm just happy it all ended well.

 

[BudWalker]

My comments were all based on the DJI Go App. I'm not sure how the Litchi app behaves. But, all that data will be in the .DAT file. If you could retrieve and post the .DAT we might know more. Look here to see how to retrieve the .DAT. It'll be large so you'll have to create a Dropbox link.

 

[ElGuapo]

Funny you mention that CRCs reality.

I had a couple of similar incidents at my work's place parking structure. I have only launched two times in the back park lot at work during my lunch hour (hope my Boss does not read this post, Lol). Right before I launched those two times everything looked good in the DJI GO App, no compass errors, no gps errors, no calibration notifications. So I launched the first time and raised my altitude to about 18 ft, well it was strange that the AC would not stay in place, in fact it started to drift away and there hardly any wind, so I tried to bring the bird back but the minute it came back to me and relaxed the controls the bird started to drift away and my control would change from P-GPS to ATTI.

Honestly, since I am new to this I thought that I needed to perform compass and IMU calibration. So went home that day and performed both calibrations, next day I went to a park close to where I live and the bird flew like a charm. That afternoon, after the calibrations, I flew close an altitude of 350 ft and about 1500 ft distance, photography and videos were nice. No issues controlling the bird during my flight, I have two batteries and both flights were otherwise uneventful . During my flight I was thinking about the first incident and thought that may be the bird needed calibration.

A few days later, during my lunch hour, I launched the bird for the second time (same spot) in the company's back parking lot and same situation occurred. Let me mention that there is a metal gate in the truck dock, and metal fencing around. There has to be some sore of strong inference going on because it is one place that I have a hard time controlling my bird.
This is the reason it is recommended in some videos not to initiate a flight where there is a lot of metal around the launch point.

 

[BudWalker]

There may be some confusion. The midflight "compass error"s are being caused by "YAW_ERROR_LARGE". These get reported in the DJI Go App as a compass error. The COMPASS_ERROR_LARGE usually happens just after launch and it also gets reported by the DJI Go App as a compass error. My first post was answering @neven 's question about compass errors at launch.

YAW_ERROR_LARGE errors often follow the COMPASS_ERROR_LARGE errors. But, they can occur otherwise. Sometimes the cause can be determined. Lately it seems that the cause is becoming more indeterminate.

 

[shockwave199]

Oh, latest firmware on the bird, batteries, and controller if that matters.
Also, I was using Litchi (latest version) for the flight, but doubt that was a factor.

I missed this info about litchi. When I had my compass error I too was using litchi and received no warning. My error happened much lower at around 75'. My only indication there was a problem was the sound of the bird flying away to my left. I was in the process of opening up a waypoint mission during what I thought was a hover. There was no warning in litchi. Upon landing, like you in atti mode behavior (not having switched to atti mode myself) I opened the dji app to see compass error in red. What I am most spooked about is the failure of litchi to ever warn me of the problem. I mentioned it in the Google group and support wasn't really too specific about an answer. I think they believe errors should be reported in the app but I'm not sold they know for sure they will all be. Like you, HD showed nothing of the incident. So, I use Litchi with caution. I absolutely do not use only the litchi app for any flight because of this. I've lost confidence in its error reporting. Each flight I start in the dji app, get airborne in that app, and confirm stability in that app. Then I swipe closed and go into litchi to continue. If I'm just flying around, I've only flown litchi a couple times to check out fpv (ios version catch up) but I much prefer the dji app, actually. I like Litchi a ton for waypoints and orbit, etc. But when the chips are really down for critical must-film situations where I must fly manually, I find much prefer the dji app. I knew that would be the case all along for me. I use Litchi and like it but I believe that overall the dji app is best for overall flying. But when you're certain the bird is stable and all is well, litchi is awesome too for what it offers.