Fixing DJI's Compass Problem

[Cor Brink]

I'm a little bit at a loss here. What does the compass do and what does the GPS do when you're flying?
My logic tells me that the compass takes care of direction (when static) while the GPS takes care of it's location and direction in space when flying.
I don't understand what the declination has to do with any of this. If I'm flying in an area where the declination is 10 degrees it cannot suddenly change. There is also a constant compass error because the compass error cannot change.
This TBE could be a function of motor response times & gains while the flight controller tries to keep the craft in one position.
Fortunately I don't have any of the 'funny' issues described on this forum that just magically happens, flyaways, TBE, who knows what else.

 

[sar104]

I'm a little bit at a loss here. What does the compass do and what does the GPS do when you're flying?
My logic tells me that the compass takes care of direction (when static) while the GPS takes care of it's location and direction in space when flying.
I don't understand what the declination has to do with any of this. If I'm flying in an area where the declination is 10 degrees it cannot suddenly change. There is also a constant compass error because the compass error cannot change.
This TBE could be a function of motor response times & gains while the flight controller tries to keep the craft in one position.
Fortunately I don't have any of the 'funny' issues described on this forum that just magically happens, flyaways, TBE, who knows what else.

The compass tells the device which way it is oriented (true bearing) by adding assumed declination to measured magnetic bearing, while the GPS tells it location and track. In GPS mode it expects actual direction of travel (GPS track) to be consistent with compass bearing and stick input, otherwise it makes roll or yaw flight corrections. If its compass bearing is wrong then it will make inappropriate corrections that will still not result in bearing and track consistency, leading to continually curved flight paths.

 

[Cor Brink]

There is no need to add or subtract declination so I doubt whether that is done.
Your explanation also explains that this has nothing to do with so-called TBE.

 

[Bigbells]

Fortunately I don't have any of the 'funny' issues described on this forum that just magically happens, flyaways, TBE, who knows what else.

There are two types of Phantom pilots: those who have had issues and those who have not. The latter tend to be overly complimentary of their own skills, right up until the point that they join the ranks of the former.

 

[Cor Brink]

Fortunately I didn't compliment my skills nor do I tend to be aloof. But maybe you talk of experience Bigballs.

 

[Bigbells]

Fortunately I didn't compliment my skills nor do I tend to be aloof. But maybe you talk of experience Bigballs.

You have referred to the problems of others in words that indicate you believe their problems are "magical". I guess you're retracting that now, and you're just saying that you have been fortunate not to have yet experienced such problems, right?

 

[sar104]

There is no need to add or subtract declination so I doubt whether that is done.
Your explanation also explains that this has nothing to do with so-called TBE.

No - declination has to be included because the GPS system uses true bearing, not magnetic. Otherwise track and heading will disagree and the NAZA will attempt to correct. Why do you think that would not matter?

TBE is almost certainly directly related, and will arise from a feedback loop created by an initial perturbation in GPS location. The NAZA will detect a drift from position (via GPS) and will apply a correction in what it thinks is the correct direction, based on its orientation (determined by the compass). If that direction is incorrect (due to misapplication of declination or a miscalibrated compass), then it will not return to the desired position, but miss it by a small amount. That will result in another attempt to correct from an already offset position, and will miss by slightly more. If the applied corrections are assumed to be parallel to (in direction) and proportional to (in magnitude) the assumed vector to the target position, and that vector is always correct in size but offset in direction by a fixed amount (the declination error) then the result (which is perfectly calculable given assumptions about the position detection accuracy and correction damping), is the device circling the target location at a radius dependent only on the declination error.

 

[ianwood]

There is no need to add or subtract declination so I doubt whether that is done.
Your explanation also explains that this has nothing to do with so-called TBE.

The adding or subtracting of magnetic declination to navigate by try true heading has been a fundamental step in basic navigation for quite some time.

No - declination has to be included because the GPS system uses true bearing, not magnetic. Otherwise track and heading will disagree and the NAZA will attempt to correct. Why do you think that would not matter?

TBE is almost certainly directly related, and will arise from a feedback loop created by an initial perturbation in GPS location. The NAZA will detect a drift from position (via GPS) and will apply a correction in what it thinks is the correct direction, based on its orientation (determined by the compass). If that direction is incorrect (due to misapplication of declination or a miscalibrated compass), then it will not return to the desired position, but miss it by a small amount. That will result in another attempt to correct from an already offset position, and will miss by slightly more. If the applied corrections are assumed to be parallel to (in direction) and proportional to (in magnitude) the assumed vector to the target position, and that vector is always correct in size but offset in direction by a fixed amount (the declination error) then the result (which is perfectly calculable given assumptions about the position detection accuracy and correction damping), is the device circling the target location at a radius dependent only on the declination error.

Great explanation.

 

[Boomerang]

I'm on the coast near the South Carolina / Georgia border. I've had my Phantom 2 Vision just over a month. For the first few weeks it was perfect; hovered in place and flew fine. Dozens of flights.

In the past week or so I've been having issues I did not have before. Some TBE and it will not hover in place even with 11 satellites locked. I have to give it lots of forward stick to keep it in place. If I don't give it constant forward stick, it flies backwards.

 

[Yeadonites]

How about starting a Facebook group or Twitter feed?
DJI are sure to react to a social media campaign.
You might also consider uploading partial serial numbers of the affected units to 'prove' the campaign is real to the naysayers.

 

[ianwood]

How about starting a Facebook group or Twitter feed?
DJI are sure to react to a social media campaign.
You might also consider uploading partial serial numbers of the affected units to 'prove' the campaign is real to the naysayers.

I've got a list of people, their location, declination and a summary of what they've seen. As for social media, it'll be a coordinated series of cluster bombs where we get as many people as we can to post the same thing on the same day to Twitter. Then we go quiet for a day or two and then we all post to FB with some photos depicting Phantom carnage. Then Youtube with one well done video that we all post to our own accounts separately at the same time.

BUT, not yet. DJI must be given time to address us first. They can make everyone happy just by saying "We hear you and we're on it." Hint, hint, DJI. Tick tock...

 

[Almmohd]

:O ooh diabolical [GRINNING FACE WITH SMILING EYES]


Sent from my iPhone using Tapatalk

 

[Cor Brink]

There is no need to add or subtract declination so I doubt whether that is done.
Your explanation also explains that this has nothing to do with so-called TBE.

No - declination has to be included because the GPS system uses true bearing, not magnetic. Otherwise track and heading will disagree and the NAZA will attempt to correct. Why do you think that would not matter?

TBE is almost certainly directly related, and will arise from a feedback loop created by an initial perturbation in GPS location. The NAZA will detect a drift from position (via GPS) and will apply a correction in what it thinks is the correct direction, based on its orientation (determined by the compass). If that direction is incorrect (due to misapplication of declination or a miscalibrated compass), then it will not return to the desired position, but miss it by a small amount. That will result in another attempt to correct from an already offset position, and will miss by slightly more. If the applied corrections are assumed to be parallel to (in direction) and proportional to (in magnitude) the assumed vector to the target position, and that vector is always correct in size but offset in direction by a fixed amount (the declination error) then the result (which is perfectly calculable given assumptions about the position detection accuracy and correction damping), is the device circling the target location at a radius dependent only on the declination error.

You tried quite hard but doesn't make sense. Track and bearing will always disagree here unless there is absolutely no wind. If what you say is like that which I doubt the drone will go forward and backwards and not in circles. I will start with big adjustments and become smaller and definitely not the other way around.

 

[sar104]

There is no need to add or subtract declination so I doubt whether that is done.
Your explanation also explains that this has nothing to do with so-called TBE.

No - declination has to be included because the GPS system uses true bearing, not magnetic. Otherwise track and heading will disagree and the NAZA will attempt to correct. Why do you think that would not matter?

TBE is almost certainly directly related, and will arise from a feedback loop created by an initial perturbation in GPS location. The NAZA will detect a drift from position (via GPS) and will apply a correction in what it thinks is the correct direction, based on its orientation (determined by the compass). If that direction is incorrect (due to misapplication of declination or a miscalibrated compass), then it will not return to the desired position, but miss it by a small amount. That will result in another attempt to correct from an already offset position, and will miss by slightly more. If the applied corrections are assumed to be parallel to (in direction) and proportional to (in magnitude) the assumed vector to the target position, and that vector is always correct in size but offset in direction by a fixed amount (the declination error) then the result (which is perfectly calculable given assumptions about the position detection accuracy and correction damping), is the device circling the target location at a radius dependent only on the declination error.

You tried quite hard but doesn't make sense. Track and bearing will always disagree here unless there is absolutely no wind. If what you say is like that which I doubt the drone will go forward and backwards and not in circles. I will start with big adjustments and become smaller and definitely not the other way around.

A wind field differs in that it is approximately constant in absolute direction, whereas the declination error is constant relative to the orientation of the device. They are indistinguishable only in linear motion in one direction - that was discussed earlier in the thread. I've no idea where you are getting your "backwards and forwards" idea from. Whenever acceleration and velocity vectors are not colinear the result is a curved path, not a linear oscillation - that is a trivially obvious result.

 

[Cor Brink]

I'm referring to the so-called TBE where it will hover forwards and backwards and not in circles.
There is no issues with linear flight.

 

[BigBadFun]

I'm referring to the so-called TBE where it will hover forwards and backwards and not in circles.

A Toilet Bowl Effect that does not go in circles?? Mate, do you actually own a Phantom and have this issue, or just here to troll?

 

[Cor Brink]

I'm referring to the so-called TBE where it will hover forwards and backwards and not in circles.

A Toilet Bowl Effect that does not go in circles?? Mate, do you actually own a Phantom and have this issue, or just here to troll?

You should probably read the previous posts before posting sarcastic no-sense, no value added comments, Mate.

 

[ianwood]

You should probably read the previous posts before posting sarcastic no-sense, no value added comments, Mate.

Let's just let it go, OK. You're not exactly adding much value either. If you have constructive thoughts on the declination issue and you actually know something about how it works like Sar104 does, by all means contribute. But knowing what declination is and why it is necessary for GPS/compass based navigation is part of the requisite table stakes for that discussion.

And as always for anyone who's experienced the issue described in the first post, check-in on this thread if you have not done so already.

 

[Cor Brink]

Being an airline pilot makes me question the things you all say here.
How many phantoms do you think have been sold and how many complain? Of the 200 members on this forum there's less than a quarter that has compass issues.
Do the maths and you'll probably realise there's more finger trouble amongst the complainers than anything else.
These are probably also the same people that 'root' and 'jailbreak' their phones and wonder why it's broken.
You're farting against thunder.

 

[terry.one]

I've requested beta testers for this issue here:
viewtopic.php?f=19&t=11555

Thanks very much for ianwood and everyone's input here. Hope to find out cause and fix it soon.