Veering right at higher altitudes..

[sar104]

Wow, thanks for the in-depth reply. Although I'm not entirely sure what it all means. So, should I calibrate EMU again and the compass?

I live over a garage, and the garage is full of scrap metal, power tools and goodness knows what else. Should I maybe calibrate the EMU somewhere else?

I've noticed, when checking camera settings inside, that I get severe compass warnings from the drone when in my living room, above the aforementioned garage. Could there be something in the garage that screws up my EMU calibration? Also, where I normally calibrate my compass is about 5 meters (outside) from the garage. Could this be causing the issue?

PS. Also, there is a train track running to the left of where I live with overhead power lines. Could this have something to do with it? "To the left" would be to the left of where the drone did the J manoeuvre first on the flight path picture you posted above. I also noticed yesterday that the drone flew pretty much bang straight at an alt of about 5 metres at full p-mode speed until it neared the tracks. Today I will take it to a location further away from the tracks and see if I still get the J-hook pattern.

View attachment 43433

The black line above shows roughly where the tracks lie in relation to the flight path.

PPS: Firmware is up to date to current if that has anything to do with it.

None of that sounds ideal, especially for compass calibration. The IMU only needs a stable, level surface. I would first repeat the compass calibration somewhere much further from any possible sources of magnetic interference. And there is certainly no harm checking whether the behavior is location dependent.

 

[BudWalker]

@jmark
If you haven't already done a compass calibration could do a test first. Don't need to fly it. Go to a location that is free of magnetic interference - not over that garage. Power up. Then hold the Mavic at arms length and turn through 3 rotations at about 10 secs per rotation. That's you rotate, i.e. don't rotate the Mavic in your hands. Try to keep the Mavic as level as possible. Then provide that .DAT.

Make sure the Go App is up and running and connected to the RC before powering on the Mavic.

 

[BudWalker]

@jmark
Have you attached something to the Mavic like a tracker?

 

[jmark]

@jmark
Have you attached something to the Mavic like a tracker?

No, I haven't attached anything to it. It's barely a week old.

@jmark
If you haven't already done a compass calibration could do a test first. Don't need to fly it. Go to a location that is free of magnetic interference - not over that garage. Power up. Then hold the Mavic at arms length and turn through 3 rotations at about 10 secs per rotation. That's you rotate, i.e. don't rotate the Mavic in your hands. Try to keep the Mavic as level as possible. Then provide that .DAT.

Make sure the Go App is up and running and connected to the RC before powering on the Mavic.

So, I will not be doing a compass calibration, but doing this test and then sending the DAT?

 

[BudWalker]

No, I haven't attached anything to it. It's barely a week old.



So, I will not be doing a compass calibration, but doing this test and then sending the DAT?

Yes, please

 

[jmark]

@BudWalker I believe I have done as requested and that this is the correct file.

18-08-09-06-16-29_FLY068.DAT

I think this is what you are asking for. I went out into the middle of the field near my house, turned on the controller, opened up the DJI app, turned on the Mavic. I then proceeded, much to the amusement of my neighbour, to turn around 3 times (about 10 seconds per turn) with the Mavic stretched out in my right hand. Was that correct?

 

[BudWalker]

@BudWalker I believe I have done as requested and that this is the correct file.

18-08-09-06-16-29_FLY068.DAT

I think this is what you are asking for. I went out into the middle of the field near my house, turned on the controller, opened up the DJI app, turned on the Mavic. I then proceeded, much to the amusement of my neighbour, to turn around 3 times (about 10 seconds per turn) with the Mavic stretched out in my right hand. Was that correct?

Well, that certainly looks like a compass with a flawed calibration. The magYaw/Yaw separation is heading dependent. The maxima are correlated with SE to S headings and the minima are correlated with N to NE headings

The gradually increasing magYaw/Yaw may be due to the FC attempting to correct the Yaw with the magYaw.

What do you think @sar104 ?

It would be really interesting for you to calibrate the compass at a better location, like where you did this test. Then without powering down do this same test.

Is there anymore to be said about the last compass calibration besides the location? Did it complete normally?

 

[jmark]

Well, that certainly looks like a compass with a flawed calibration. The magYaw/Yaw separation is heading dependent. The maxima are correlated with SE to S headings and the minima are correlated with N to NE headings
View attachment 43447
The gradually increasing magYaw/Yaw may be due to the FC attempting to correct the Yaw with the magYaw.

What do you think @sar104 ?

It would be really interesting for you to calibrate the compass at a better location, like where you did this test. Then without powering down do this same test.

Is there anymore to be said about the last compass calibration besides the location? Did it complete normally?

Well, I'm literally blown away by the help I'm getting on this forum. Thank you very much guys.

I'll try and do a new compass calibration tomorrow, weather permitting, in the middle of the field where I did this test. And then send the results.

Thanks again!

 

[sar104]

@BudWalker - this looks potentially like another case of IMU errors causing bad compass data, but with only one IMU it's harder to identify.

Well, that certainly looks like a compass with a flawed calibration. The magYaw/Yaw separation is heading dependent. The maxima are correlated with SE to S headings and the minima are correlated with N to NE headings
View attachment 43447
The gradually increasing magYaw/Yaw may be due to the FC attempting to correct the Yaw with the magYaw.

What do you think @sar104 ?

It would be really interesting for you to calibrate the compass at a better location, like where you did this test. Then without powering down do this same test.

Is there anymore to be said about the last compass calibration besides the location? Did it complete normally?

Agreed.

 

[jmark]

@BudWalker - this looks potentially like another case of IMU errors causing bad compass data, but with only one IMU it's harder to identify.


Agreed.

So, this is not likely a warranty issue? And more just an issue with how I have been calibrating the drone?

 

[BudWalker]

So, this is not likely a warranty issue? And more just an issue with how I have been calibrating the drone?

Actually, my opinion is that it's not, or shouldn't be, possible to obtain a flawed compass calibration. There are a lot of people who don't agree though. Let's wait and see what the results are from your compass calibration.

 

[jmark]

Okay guys, went out and did as requested this morning…

18-08-11-09-35-52_FLY073.DAT

Went out in to the field where the other test was performed. Calibrated the compass and without turning off the craft did 3x 10 second turns.

Hope this was correct and we can finally get to the bottom of this. If not, I’ve decided to send the craft in on its warranty on Monday. Which means, that I’ll be without my drone for at least 2 week to a month. It’s a shame, the craft is otherwise superb and I can see it’s extreme potential for my photography and filming. So, lets hope it doesn’t come to that. Fingers crossed.

In addition. I took it for a quick test flight after the calibration and test. I had turned the device off to set up my takeoff point, but still had the same issue. Here are the files of the test flight...

Flight Log

DJIFlightRecord_2018-08-11_[09-53-14].txt

DAT log

18-08-11-09-52-52_FLY074.DAT

I did several back and forth flights in different directions, sometimes the veer was more, sometimes less.

 

[sar104]

Okay guys, went out and did as requested this morning…

18-08-11-09-35-52_FLY073.DAT

Went out in to the field where the other test was performed. Calibrated the compass and without turning off the craft did 3x 10 second turns.

Hope this was correct and we can finally get to the bottom of this. If not, I’ve decided to send the craft in on its warranty on Monday. Which means, that I’ll be without my drone for at least 2 week to a month. It’s a shame, the craft is otherwise superb and I can see it’s extreme potential for my photography and filming. So, lets hope it doesn’t come to that. Fingers crossed.

In addition. I took it for a quick test flight after the calibration and test. I had turned the device off to set up my takeoff point, but still had the same issue. Here are the files of the test flight...

Flight Log

DJIFlightRecord_2018-08-11_[09-53-14].txt

DAT log

18-08-11-09-52-52_FLY074.DAT

I did several back and forth flights in different directions, sometimes the veer was more, sometimes less.

That doesn't look any better - similar variation as before:



It's clearer to see in the yaw domain:



The maximum disagreement is still at around -25° (NNW).

The issue can be seen in the actual magnetometer data, but it's still hard to figure out the direct cause. The magZ data should be approximately constant, but instead show significant variation with almost a discontinuity around the point of maximum variation.



Have you also done an IMU calibration?

 

[jmark]

That doesn't look any better - similar variation as before:

View attachment 43526

It's clearer to see in the yaw domain:

View attachment 43527

The maximum disagreement is still at around -25° (NNW).

The issue can be seen in the actual magnetometer data, but it's still hard to figure out the direct cause. The magZ data should be approximately constant, but instead show significant variation with almost a discontinuity around the point of maximum variation.

View attachment 43532

Have you also done an IMU calibration?

I have not done one today, although I have performed several IMU calibrations this week. It has not helped.

I guess, the only alternative is to either send it back to the store or send it in on it's warranty?

 

[BudWalker]

Okay guys, went out and did as requested this morning…

18-08-11-09-35-52_FLY073.DAT

Went out in to the field where the other test was performed. Calibrated the compass and without turning off the craft did 3x 10 second turns.

Hope this was correct and we can finally get to the bottom of this. If not, I’ve decided to send the craft in on its warranty on Monday. Which means, that I’ll be without my drone for at least 2 week to a month. It’s a shame, the craft is otherwise superb and I can see it’s extreme potential for my photography and filming. So, lets hope it doesn’t come to that. Fingers crossed.

In addition. I took it for a quick test flight after the calibration and test. I had turned the device off to set up my takeoff point, but still had the same issue. Here are the files of the test flight...

Flight Log

DJIFlightRecord_2018-08-11_[09-53-14].txt

DAT log

18-08-11-09-52-52_FLY074.DAT

I did several back and forth flights in different directions, sometimes the veer was more, sometimes less.

The calibration didn't seem to make any difference.

Before you send it off for it's 2 - 4 week journey through DJI repair consider this. DJI may just say that it operates within specs and there isn't anything to fix. Do you not have the option of returning it for a replacement?

The flight seem to be straight except at the beginning and the end. For this section of the latest test flight


The Mavic starts with the directionOfFlight veering from Yaw by about 15°. But after about 6 secs the directionOfFlight is aligned with Yaw.


That's excessive. I tried mine this morning and it also does this but to a much lesser degree. Only 5° but then aligns itself in about 0.5 secs.

Is it possible that you could change your operating procedure and start the high speed run 6 or 10 secs earlier.?

From watching your video part of the problem may be that, in addition to not flying straight, the gimbal is pointed off to the side. I checked this by looking at the gimbalYaw in the .txt compared to the Yaw in the .DAT. The difference was less that 1°. Still, you might want to calibrate the gimbal.

As I'm writing I'm seeing that @sar104 suggested that you do an IMU calibration. I agree, but am not hopeful. I hadn't noticed the magZ discontinuity. Typically, mag(1) is less well behaved than mag(0) and mag(0) is the default. Maybe mag(1) is being used. The tablet .DAT doesn't allow us to know which compass is being used. We need the AC .DAT for that.

 

[jmark]

The calibration didn't seem to make any difference.

Before you send it off for it's 2 - 4 week journey through DJI repair consider this. DJI may just say that it operates within specs and there isn't anything to fix. Do you not have the option of returning it for a replacement?

The flight seem to be straight except at the beginning and the end. For this section of the latest test flight

View attachment 43530
The Mavic starts with the directionOfFlight veering from Yaw by about 15°. But after about 6 secs the directionOfFlight is aligned with Yaw.
View attachment 43531

That's excessive. I tried mine this morning and it also does this but to a much lesser degree. Only 5° but then aligns itself in about 0.5 secs.

Is it possible that you could change your operating procedure and start the high speed run 6 or 10 secs earlier.?

From watching your video part of the problem may be that, in addition to not flying straight, the gimbal is pointed off to the side. I checked this by looking at the gimbalYaw in the .txt compared to the Yaw in the .DAT. The difference was less that 1°. Still, you might want to calibrate the gimbal.

As I'm writing I'm seeing that @sar104 suggested that you do an IMU calibration. I agree, but am not hopeful. I hadn't noticed the magZ discontinuity. Typically, mag(1) is less well behaved than mag(0) and mag(0) is the default. Maybe mag(1) is being used. The tablet .DAT doesn't allow us to know which compass is being used. We need the AC .DAT for that.

To be honest, knowing how companies operate over here, if I ask for a replacement they may well just tell me it's a warranty issue and it's nothing to do with them any more. I will try though. The drone was bought with a 30 day money back guarantee, so I may just use that instead. Going through the hassle of a warranty claim and waiting up to 4 weeks for a device that should be flawless at that price, is not something I think I should be expected to put up with. And as well, like you said, DJI may just brush it off.

Thank you so much for your help guys. You've been nothing short of amazing. I really appreciate it.

 

[Cheech Wizard]

The IMU is very finicky, you have to have a totally level platform to do this on. When you calibrate the compass, are you using the drone as a center point or your body? In other words are you holding the drone out and spinning your body or walking your body around the drone? And as stated by other posters, get far away from the scrape metal and railroad tracks when you calibrate the compass. You can and should do the IMU inside on a PERFECTLY level surface.

 

[jmark]

The IMU is very finicky, you have to have a totally level platform to do this on. When you calibrate the compass, are you using the drone as a center point or your body? In other words are you holding the drone out and spinning your body or walking your body around the drone? And as stated by other posters, get far away from the scrape metal and railroad tracks when you calibrate the compass. You can and should do the IMU inside on a PERFECTLY level surface.

When I calibrate the compass I hold the drone out at arms length and spin around in a circle slowly. The drone isn't the centre point, my body is, and the drone sort of orbits around my body, if you understand what I mean?

 

[Cheech Wizard]

When I calibrate the compass I hold the drone out at arms length and spin around in a circle slowly. The drone isn't the centre point, my body is, and the drone sort of orbits around my body, if you understand what I mean?

You need to try it the other way around, you want a tight circle on the drone.. give it a shot. I hold the drone down, put my head over it and spin my body around slowly, tilt it and do it again, and yes my neighbors think I'm weird too. There are different opinions on this, so get ready.

Also saw on a post that a smart watch (or other device) could affect the magnetic field.

 

[jmark]

Hey everyone. Just an update. I ahev decided to send the drone back on the stores 30 day try before you buy policy. With the likelihood of the mavic 2 release being soon, I'll wait for that instead.

Just want to say thank you for all the help I received. You guys are awesome!