Fixing DJI's Compass Problem

[jimre]

I thought the point of doing the compass calibration "dance" was to adjust for local declination. Is that not the case?

You need to read the very first post.

I read all the posts here, thank you. The first posts lays out the hypothesis of the original poster quite nicely. But it's just a hypothesis. I would concur there is likely a bug of some kind here - but I also assume the DJI engineers are very smart people, and not likely to just "forget" something so basic - which this thread is clearly accusing them of.

I'd like to understand what is *supposed* to be happening with the "compass dance", that's all. I had assumed it was to take local declination and/or interference into account.

 

[aussiepilot]

I get the same problem, have flown in far North Queensland and Sydney on easyt coast of Australia.
My p2v hooks left going forward and right going back. It also has toilet bowl effect in hover.

 

[sar104]

I thought the point of doing the compass calibration "dance" was to adjust for local declination. Is that not the case?

You need to read the very first post.

I read all the posts here, thank you. The first posts lays out the hypothesis of the original poster quite nicely. But it's just a hypothesis. I would concur there is likely a bug of some kind here - but I also assume the DJI engineers are very smart people, and not likely to just "forget" something so basic - which this thread is clearly accusing them of.

I'd like to understand what is *supposed* to be happening with the "compass dance", that's all. I had assumed it was to take local declination and/or interference into account.

No - it has nothing to do with declination. Rotating the compass simply exposes the sensor to the full 3-D local magnetic field so that it can then estimate an accurate 2-D magnetic bearing from an individual resolved field measurement. It provides no information on the relationship between magnetic north and the arbitrary true north except via the calculated (or via lookup table) declination.

 

[jimre]

I'd like to understand what is *supposed* to be happening with the "compass dance", that's all. I had assumed it was to take local declination and/or interference into account.

No - it has nothing to do with declination. Rotating the compass simply exposes the sensor to the full 3-D local magnetic field so that it can then estimate an accurate 2-D magnetic bearing from an individual resolved field measurement. It provides no information on the relationship between magnetic north and the arbitrary true north except via the calculated (or via lookup table) declination.

Thank you. So then the primary purpose of the "dance" is to identify & compensate for local magnetic interference?

 

[ianwood]

Thank you. So then the primary purpose of the "dance" is to identify & compensate for local magnetic interference?

Sort of. It is measuring the magnetic field in 3 axis to get a baseline reading which will vary by region and based on what you have strapped onto your Phantom. There would be no way for the compass to know that magnetic north is xx degrees from true north given that true north is like the equator, a made up point of reference that cannot be detected magnetically or otherwise.

 

[droneranger]

Excellent evaluation of the problem ianwood. I have been experiencing the exact same symptoms that you describe and show in your video. I live in the City of Ventura in Southern California and have tried all sorts of calibration "dances" and rotational procedures to no effect. Just as you say, the issue seems to damp out and lessen after a few minutes of flight time, but then reappears on the next power-up. VERY FRUSTRATING!! I had a P2V before purchasing a P2/Zenmuse unit, and it did not exhibit this behavior. Thanks for taking the lead on this. Hopefully DJI will address this sooner rather than later. I will keep looking at this thread for updates.

BTW, Dronefly is in Newbury Park, which is around 20 minutes from my home. I don't have any "pull" there, but they do know me as a customer if that is of any help.

 

[jimre]

Thank you. So then the primary purpose of the "dance" is to identify & compensate for local magnetic interference?

Sort of. It is measuring the magnetic field in 3 axis to get a baseline reading which will vary by region and based on what you have strapped onto your Phantom. There would be no way for the compass to know that magnetic north is xx degrees from true north given that true north is like the equator, a made up point of reference that cannot be detected magnetically or otherwise.

I would imagine the way that declination is "detected" would be some type of lookup based on current GPS coordinates. This is one of the things that I had assumed was happening during the compass dance - and why the dance only needed to be repeated if you had changed location significantly since the last flight.

 

[sar104]

Thank you. So then the primary purpose of the "dance" is to identify & compensate for local magnetic interference?

Sort of. It is measuring the magnetic field in 3 axis to get a baseline reading which will vary by region and based on what you have strapped onto your Phantom. There would be no way for the compass to know that magnetic north is xx degrees from true north given that true north is like the equator, a made up point of reference that cannot be detected magnetically or otherwise.

I would imagine the way that declination is "detected" would be some type of lookup based on current GPS coordinates. This is one of the things that I had assumed was happening during the compass dance - and why the dance only needed to be repeated if you had changed location significantly since the last flight.

These are still two quite separate issues. The declination is purely a function of location - provided by the GPS unit. It does not need the compass calibration dance to get that information - it uses the GPS location together with a lookup table or standard model. The calibration is needed because the magnetic field varies with location in more ways than just the tangential field component direction - the radial component also varies and has to be subtracted to give the magnetic bearing.

 

[flight-of-eye]

Has this been sent to DJI yet?

 

[ianwood]

Has this been sent to DJI yet?

I answered a similar question in the other thread:

any reaction from DJI yet?

Only through entry-level contacts so far and the response there has been predictably dismissive. Paraphrasing but the response I received was "68 people is nothing compared to thousands of units sold" and "we appreciate your concern", "send your Phantom in for an exchange", etc.

This is exactly as I expected which is why I am in the process of identifying higher-level DJI staff who actually have some real contact with the FC development team. Obviously, DJI cannot guarantee a fix nor can they really formally acknowledge the problem from a liability perspective. That's fine. But I wholeheartedly expect an unofficial "don't you worry, we're on it" that I can relay to you guys so we can all take some comfort in knowing it's getting the right attention it needs.

If the mid-level contacts are equally dismissive, then we escalate. There is a "nuclear option" but let's exhaust all diplomatic options first. We have to give DJI the full opportunity to reassure us that they're taking the problem seriously.

 

[SlackerATX]

How much does this affect your ability to yaw in place after you first power on?

 

[flight-of-eye]

My concern is with the newley available app update coming. Will we have waypoint issues and catastrophic flyways due to this issue. I certainly do not want to send it on its way without knowing if it will be arriving at the point intended. I can see this as a sure thing, "Oh wait, it just went through the window of the building on the left at destination 2. I swear officer it was supposed to go straight."

 

[Sjiekong]

My Phantom 2 doesn't hover steady, doesn't keep it's height, even after successful calibration. Would you consider it the same issue?

 

[flyNfrank]

ianwood, what you are doing is commendable. I appreciate your efforts and dedication to seeing things through. With that said, I would like to see you spend just a little more time explaining things. Please don't think I'm knocking any of what you have done so far. I think you should explain the problem as if you were doing it in front of people you know for a fact that do not know what a Multicopter is. Again don't get me wrong I followed you right along but several things came to. mind in the process.

Maybe this will help with what I'm kind of referring to. Your explaination was all oral. I would have liked seeing you start out draw it out on paper as you were describing it. In doing this you can point out how little or how bad it is doing what you are about to show in the video. Then I can better focus on what is going on rather than looking around at other stuff on the screen I'm looking at. Even blurring out all that's visible on the sides would have been a good thing. And another would been to try and capture you working the controller from behind while being able to see the P2 fly away in it's hook. That takes care of the guys that says "How do we know he is pressing 100% forward, blah, blah...

I personally would have like to see you in the very beginning do a Compass Calibration, and also run the controller a stick calibration. That way all would know that part of the P2 was done, and done correctly before hand. Doing these things show me that much more how serious you are at finding or getting the support to find a solution to the problem.

Now I know anyone can sit back and complain about how somebody does something. But I want you to know I'm only pointing out to help the cause. You might think the stuff I mentioned is overkill for what you're doing. But I think it is right on par with what needs to be done for everyone to see this is a for real deal, so to speak.

Whatever you do, keep doing what you're doing. I luv it!

 

[ianwood]

How much does this affect your ability to yaw in place after you first power on?

It doesn't. At least not for me. Users in areas with very high declination (+/-18) may get some TBE right at takeoff but otherwise stability and yawing on take off are not compromised at all.

My Phantom 2 doesn't hover steady, doesn't keep it's height, even after successful calibration. Would you consider it the same issue?

This sounds like a different issue.

ianwood, what you are doing is commendable. I appreciate your efforts and dedication to seeing things through. With that said, I would like to see you spend just a little more time explaining things. Please don't think I'm knocking any of what you have done so far. I think you should explain the problem as if you were doing it in front of people you know for a fact that do not know what a Multicopter is. Again don't get me wrong I followed you right along but several things came to. mind in the process.

Maybe this will help with what I'm kind of referring to. Your explaination was all oral. I would have liked seeing you start out draw it out on paper as you were describing it. In doing this you can point out how little or how bad it is doing what you are about to show in the video. Then I can better focus on what is going on rather than looking around at other stuff on the screen I'm looking at. Even blurring out all that's visible on the sides would have been a good thing. And another would been to try and capture you working the controller from behind while being able to see the P2 fly away in it's hook. That takes care of the guys that says "How do we know he is pressing 100% forward, blah, blah...

I personally would have like to see you in the very beginning do a Compass Calibration, and also run the controller a stick calibration. That way all would know that part of the P2 was done, and done correctly before hand. Doing these things show me that much more how serious you are at finding or getting the support to find a solution to the problem.

Now I know anyone can sit back and complain about how somebody does something. But I want you to know I'm only pointing out to help the cause. You might think the stuff I mentioned is overkill for what you're doing. But I think it is right on par with what needs to be done for everyone to see this is a for real deal, so to speak.

Whatever you do, keep doing what you're doing. I luv it!

The video is not meant to be exhaustive. It is merely to demonstrate the actual problem as it is more descriptive than a picture. But if you want a picture and other details, the document below which is referenced in the first post provides quite a bit of it and is intended to be a more comprehensive summary of the situation:

DJI Phantom and Magnetic Declination

 

[wworrall]

I am in Anchorage Alaska and I have this problem big time. It gets a little better after flying for a while but never flies straight despite compass calibrations in a metalic free area. I am not happy with this. It hooks badly enough that I have to constantly put in left or right stick correction, and in hover the TBE is so bad sometimes that I have to make sure I stay well away from any obstacles.
I hope you can get them to fix this issue.

 

[droneranger]

ianwood,

If there is anything I can do to help move this forward, please let me know. I'd also love to hear what your "instant sunshine" option is. You can PM me on that if you prefer.

 

[wildwoodpugh]

Ian, are you having any success with DJI on the compass calibration problem?

 

[sprung]

i am in southern vermont. defiantly have this problem

 

[Jeff_S]

I just joined the forum. I've been lurking since December. I live in New Hampshire.

My experience is mixed, and I can't tell if I have the problem. I constantly use yaw as I'm flying to get good video and I always have a minor toilet bowl effect for the first several minutes which goes away. To simulate the issue, I hovered for a few minutes and then moved straight forward for a couple hundred feet and then straight back, however it came back about 50' to my right.

Now that it's warmer, I'll fly many more times but I'm curious how many people don't get a straight return.

PS My sticks are calibrated, my IMU is calibrated, my compass has been calibrated after I see a major toilet bowl effect, which did not occur for my comments above.