Mavic Compass /GPS Calibration Problems -Fixed?

[mereflyer]

Hi Fellow Mavicites,

I have noticed that many of our Pilots are having problems with the precision of the Compass/GPS combos with our Mavics. Often being the cause of annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal and highly likely to be a main cause in many reported cases of flyaways and crashes.

In my experience, although having not yet had my MP1 fly away, as others have complained, I have experienced some of the other problems. In particular the annoying failure to land on the spot with RTH, which was becoming worse with every second landing. (My Mavic would land up to ten feet away from the pad if I had not taken remedial action)

I found this so annoying and disappointing, especially as the little machine had been so accurate for the first 6 months of ownership.

I had regularly re-calibrated the compasses and IMU as per instructions and advice from the Members on these DJI Forums but never happy with the results.

I wondered if my actual method of calibration (and perhaps others as well) Could possibly have something to do with the inaccurate results we were getting!

Being from an Engineering and electronic background and an Inventive mind, I wondered if using a basic type of wooden turn-table (containing no metallic materials), to perform the calibrations would make any difference.

Well to cut a long story a bit shorter, I went ahead, designed and built a little prototype and "voila"-problem fixed.

Now my Mavic is back again to the precise landing and accurate handling machine as it was when new, I could not be happier now with its wonderful "spot on" accuracy.

Proves to me that the technique I (and obviously as many of you have found) this calibration thing is or can be, very sensitive and very important to perform correctly, for safety and preventing the possibility of an expensive loss of an unrecoverable flyaway.

Now I know others of you will dispute my findings, I won't argue with that, and whether or not these findings will be accepted is of course open to dispute. But I will say with complete confidence, that the difference made to my Mavic Pro, by using my rotating and tilting precise calibration turntable, has me convinced that this is a problem that can be fixed by using a device which is far more precise than the "human compass dance" we are instructed to follow in the manuals.

To be honest I don't see myself being able or interested in the actual production or sales of this very useful item, but could be persuaded to cooperate in design/ manufacture by anyone interested in running with it (for a reasonable commission or royalty etc...) What do you think?

Cheers,
Rob W (mereflyer)

 

[Steady Eddie]

Hi Fellow Mavicites,

I have noticed that many of our Pilots are having problems with the precision of the Compass/GPS combos with our Mavics. Often being the cause of annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal and highly likely to be a main cause in many reported cases of flyaways and crashes.

In my experience, although having not yet had my MP1 fly away, as others have complained, I have experienced some of the other problems. In particular the annoying failure to land on the spot with RTH, which was becoming worse with every second landing. (My Mavic would land up to ten feet away from the pad if I had not taken remedial action)

I found this so annoying and disappointing, especially as the little machine had been so accurate for the first 6 months of ownership.

I had regularly re-calibrated the compasses and IMU as per instructions and advice from the Members on these DJI Forums but never happy with the results.

I wondered if my actual method of calibration (and perhaps others as well) Could possibly have something to do with the inaccurate results we were getting!

Being from an Engineering and electronic background and an Inventive mind, I wondered if using a basic type of wooden turn-table (containing no metallic materials), to perform the calibrations would make any difference.

Well to cut a long story a bit shorter, I went ahead, designed and built a little prototype and "voila"-problem fixed.

Now my Mavic is back again to the precise landing and accurate handling machine as it was when new, I could not be happier now with its wonderful "spot on" accuracy.

Proves to me that the technique I (and obviously as many of you have found) this calibration thing is or can be, very sensitive and very important to perform correctly, for safety and preventing the possibility of an expensive loss of an unrecoverable flyaway.

Now I know others of you will dispute my findings, I won't argue with that, and whether or not these findings will be accepted is of course open to dispute. But I will say with complete confidence, that the difference made to my Mavic Pro, by using my rotating and tilting precise calibration turntable, has me convinced that this is a problem that can be fixed by using a device which is far more precise than the "human compass dance" we are instructed to follow in the manuals.

To be honest I don't see myself being able or interested in the actual production or sales of this very useful item, but could be persuaded to cooperate in design/ manufacture by anyone interested in running with it (for a reasonable commission or royalty etc...) What do you think?

Cheers,
Rob W (mereflyer)

How about a picture of your turntable/tilt device? Is it a gimbal type thing?

 

[mereflyer]

How about a picture of your turntable/tilt device? Is it a gimbal type thing?

If you are interested, PM me and I will give you some details

 

[Meta4]

Now I know others of you will dispute my findings, I won't argue with that, and whether or not these findings will be accepted is of course open to dispute. But I will say with complete confidence, that the difference made to my Mavic Pro, by using my rotating and tilting precise calibration turntable, has me convinced that this is a problem that can be fixed by using a device which is far more precise than the "human compass dance" we are instructed to follow in the manuals.

What comes across from your post is a lack of understanding of what compass calibration does (and what it can't do).
What you've done is to "fix" a problem that doesn't exist.
The ideas that compass calibration needs to be done frequently or that the way it's done makes a substantial difference aren't supported by an understanding of what compass calibration does and how it works.
Your compass calibration is most unlikely to be the cause of annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal or to be a main cause in many reported cases of flyaways and crashes.

Your compass is a direction indicator, it has no effect on how close your drone lands in an RTH autolanding.
Landing within 10 feet is within the precision you can expect from GPS and is quite normal.
You get more precise landings if you use the Precision Landing feature that uses optical technology.

The only thing that calibrating your compass does is to measure the magnetic fields that are part of the drone so they can be subtracted from the compass data and provide accurate compass information.
Once you have measured those fields, they don't change unless you modify or rebuild the drone.
There's no need to recalibrate frequently.
The compass calibrates or it doesn't.
There's no inbetween and it doesn't matter what technique you use.
If your drone is flying straight and hovers in place without slowly spiralling, the compass is calibrated as well as it can be.

 

[Nakita Mosquita]

Changing locations significantly (ie flying in Kaui versus flying in Ontario) prompts calibration. I've had my MA for one year now and remain uncertain of the 'proper' way to perform the calibration.
1. Does the direction of the rotation matter?
2. When tilting the drone on it's side, during the 2nd part of the calibration, some illustrations show the camera pointing horizontally, while some show the camera pointing down. Does it matter? Which is correct?

 

[Starz]

Changing locations significantly (ie flying in Kaui versus flying in Ontario) prompts calibration. I've had my MA for one year now and remain uncertain of the 'proper' way to perform the calibration.
1. Does the direction of the rotation matter?
2. When tilting the drone on it's side, during the 2nd part of the calibration, some illustrations show the camera pointing horizontally, while some show the camera pointing down. Does it matter? Which is correct?

Page 54 here. Actually quite a bit of info in this manual. DJI method correct? It’s the method I’m using and haven’t had any issues.
I don’t calibrate anything unless I see a problem in the sensor section of the app, which is probably the best indicator to check.
The Mavic air seems to have more compass calibration requests than I care to deal with.

https://dl.djicdn.com/downloads/mavic/20171219/Mavic%20Pro%20User%20Manual%20V2.0.pdf

 

[Meta4]

1. Does the direction of the rotation matter?
2. When tilting the drone on it's side, during the 2nd part of the calibration, some illustrations show the camera pointing horizontally, while some show the camera pointing down. Does it matter? Which is correct?

Either way works for both questions.
All that matters is that you rotate the drone a full 360 degrees on each axis.

 

[mereflyer]

What comes across from your post is a lack of understanding of what compass calibration does (and what it can't do).
What you've done is to "fix" a problem that doesn't exist.
The ideas that compass calibration needs to be done frequently or that the way it's done makes a substantial difference aren't supported by an understanding of what compass calibration does and how it works.
Your compass calibration is most unlikely to be the cause of annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal or to be a main cause in many reported cases of flyaways and crashes.

Your compass is a direction indicator, it has no effect on how close your drone lands in an RTH autolanding.
Landing within 10 feet is within the precision you can expect from GPS and is quite normal.
You get more precise landings if you use the Precision Landing feature that uses optical technology.

The only thing that calibrating your compass does is to measure the magnetic fields that are part of the drone so they can be subtracted from the compass data and provide accurate compass information.
Once you have measured those fields, they don't change unless you modify or rebuild the drone.
There's no need to recalibrate frequently.
The compass calibrates or it doesn't.
There's no inbetween and it doesn't matter what technique you use.
If your drone is flying straight and hovers in place without slowly spiralling, the compass is calibrated as well as it can be.

Thank You Meta, Your opinion valued and yes certainly you have some good and valid points there.

Because of the very subtle coordination of the twin compasses, GPS, the Magnetometer and accelerometers installed in these amazing little machines, also the incredible mix of software to make the entire process work in such a seamless manner it may well be that the interaction benefits from a more accurate calibration as I have proven by using my device.

I am thinking of how difficult it may be in some cases where the human method of calibration unavoidably mixes a combination of horizontal and vertical component into each of of the two movements, this may actually make the process ineffective or should I say less effective.

By making sure each of the two steps are performed on a completely accurate horizontal and then vertical plane, my device actually improves the process and thus the accuracy that results.

It is very possible there is some variability among our individual MP's, which would explain the differences in sensitivity, which in the case of my own Mavic made it a much more accurate machine than before - as previously noted.

It is of course possible, some of the MP's can handle a mixture of the two, or even some pilots may have a much steadier and more accurate method of rotation than others (such as in my case for instance)

Anyway all I can say to you - I have definitely noticed the improvement and happy to pass it on to any of you as my personal findings only and if any Mavic can be saved or improved by this, it will be worth the effort.

 

[Meta4]

in the case of my own Mavic made it a much more accurate machine than before - as previously noted.

I'm at a loss to see what aspect of your drone's operation could be more accurate as a result of any calibration magic you might have performed.
You mentioned annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal and highly likely to be a main cause in many reported cases of flyaways and crashes, in your original post.
The compass and the precision of its calibration has no influence on any of these things.

Anyway all I can say to you - I have definitely noticed the improvement

What improvements have you observed?
You only mentioned your RTH landing accuracy which is due to using the inbuilt Precision Landing feature and has nothing to do with the compass.
I suspect that the precise accuracy of your calibration method makes no difference to the outcome but would be interested to hear any comments from @sar104 who has a very good understanding of the fine points of DJI compass calibration.

 

[DanMan32]

Your Precision Landing accuracy may have improved, assuming you have enabled it before, is because of the high contrast of your pad. A good IMU calibration may help a little.

 

[sar104]

I'm at a loss to see what aspect of your drone's operation could be more accurate as a result of any calibration magic you might have performed.
You mentioned annoying and frustrating flight inaccuracies, failing to follow commands, innacurate RTH's, loss of signal and highly likely to be a main cause in many reported cases of flyaways and crashes, in your original post.
The compass and the precision of its calibration has no influence on any of these things.

What improvements have you observed?
You only mentioned your RTH landing accuracy which is due to using the inbuilt Precision Landing feature and has nothing to do with the compass.
I suspect that the precise accuracy of your calibration method makes no difference to the outcome but would be interested to hear any comments from @sar104 who has a very good understanding of the fine points of DJI compass calibration.

Sounds like two separate questions in arise here - (1) would a precision rotation sequence improve the compass calibration and (2) would a better compass calibration help with precision RTH or anything else.

With regard to (1) - as mentioned, the purpose of the compass calibration is to determine the magnetic field of the aircraft itself, so that during flight it can be subtracted from the measured magnetic field, leaving just the earth's magnetic field which is used to track magnetic north. Rotation of the aircraft allows it to detect what appears, in the aircraft frame of reference, to be a static offset (its own field) to the varying field field (the earth's). Rotation of the 3-axis magnetometer through 360° around a single axis doesn't achieve that because that still leaves the magnetometer on that rotation axis having only pointed in a single direction. Rotation around 2 axes, even if they are not perfectly orthogonal, provides magnetometer data on all three axes. The aircraft does need to know its orientation at all times during the calibration, which it gets from the IMU. The only reason a precision rotation mechanism could possibly help is if the operator were unable, unassisted, to rotate the aircraft smoothly enough to avoid compromising the IMU measurements. That would seem most unlikely, but I've never deliberately tried to achieve that.

Conclusion - yes, in principle it might help achieve smooth rotation but the exact axes don't matter.

With regard to (2) - a poor compass calibration leads to errors in the aircraft's detection of magnetic north. Typical calibration errors produce compass errors of no more than 10 or 15°. Larger compass errors are almost always caused by magnetic interference, generally at power up leading to incorrect initialization of the IMU yaw value, and will cause significant loss of flight control and stability. Crabbing flight is the common sign of a poor calibration. It has no effect on GPS positioning and little effect on P-mode holding or precision RTH because, as noted above, those are independent systems. If it can hover then it can precision RTH provided it has an optically recorded takeoff point with good contrast that doesn't change during the flight.

Conclusion - no - the quality of the compass calibration isn't going to affect RTH precision or cause flyaways, unresponsiveness to commands or other related problems. It might cause slight crabbing during straight flight.

 

[mereflyer]

Thank you both for your valuable opinions - there is no doubt about your expertise in these areas. Your observations are probably very correct.

In reply I will simply say again that the series of landing images I posted were not controlled manually, this was not possible before the calibrations made with the accuracy of my device.

I have been using this same landing pad for many months now - well before the idea and use of the rotating turntable, so not likely to be the main contributor to the better results I have experienced.
BTW not only a more accurate RTH, but also a smoother and more accurate flight control response - more "locked in" is the best way to describe the difference when flying the drone.

Another reason for the changes may just be explained by the more remote proximity of things we wear on our persons, such as watches, phones, belt buckles etc, all of which could and probably should be removed before doing the "dance" (much less of an influence with a turntable free of any metallic materials - easily operated at arms length!)

In any case I reiterate the reason for doing this experiment in the first place, was the disappointing inaccuracies, I and very obviously many, many other Mavic Pilots were experiencing, as we see in these forums. My results are positive and worth sharing!

In closing now, I offer this piece of advice "for what its worth" To all of you Mavic drone Pilots who are, or have experienced any annoying or inaccurate flight problems with your Mavics.

Just look at the idea of making your own simple, wooden or plastic rotating turntable and see if it improves landing accuracies, safety and the enjoyment factor for your drone flights.

If you do this, please share your findings here so we can all benefit!

 

[sar104]

Thank you both for your valuable opinions - there is no doubt about your expertise in these areas. Your observations are probably very correct.

In reply I will simply say again that the series of landing images I posted were not controlled manually, this was not possible before the calibrations made with the accuracy of my device.

I have been using this same landing pad for many months now - well before the idea and use of the rotating turntable, so not likely to be the main contributor to the better results I have experienced.
BTW not only a more accurate RTH, but also a smoother and more accurate flight control response - more "locked in" is the best way to describe the difference when flying the drone.

Another reason for the changes may just be explained by the more remote proximity of things we wear on our persons, such as watches, phones, belt buckles etc, all of which could and probably should be removed before doing the "dance" (much less of an influence with a turntable free of any metallic materials - easily operated at arms length!)

In any case I reiterate the reason for doing this experiment in the first place, was the disappointing inaccuracies, I and very obviously many, many other Mavic Pilots were experiencing, as we see in these forums. My results are positive and worth sharing!

In closing now, I offer this piece of advice "for what its worth" To all of you Mavic drone Pilots who are, or have experienced any annoying or inaccurate flight problems with your Mavics.

Just look at the idea of making your own simple, wooden or plastic rotating turntable and see if it improves landing accuracies, safety and the enjoyment factor for your drone flights.

If you do this, please share your findings here so we can all benefit!

If you want to attempt to verify whatever effect you might be seeing, then I suggest that you do a couple of manual calibrations, then a couple of your precision calibrations. The resulting DAT file will contain the compass calibration data and show, for starters, whether it is even producing a significantly different calibration.

 

[GadgetGuy]

Thank you both for your valuable opinions - there is no doubt about your expertise in these areas. Your observations are probably very correct.

In reply I will simply say again that the series of landing images I posted were not controlled manually, this was not possible before the calibrations made with the accuracy of my device.

I have been using this same landing pad for many months now - well before the idea and use of the rotating turntable, so not likely to be the main contributor to the better results I have experienced.
BTW not only a more accurate RTH, but also a smoother and more accurate flight control response - more "locked in" is the best way to describe the difference when flying the drone.

Another reason for the changes may just be explained by the more remote proximity of things we wear on our persons, such as watches, phones, belt buckles etc, all of which could and probably should be removed before doing the "dance" (much less of an influence with a turntable free of any metallic materials - easily operated at arms length!)

In any case I reiterate the reason for doing this experiment in the first place, was the disappointing inaccuracies, I and very obviously many, many other Mavic Pilots were experiencing, as we see in these forums. My results are positive and worth sharing!

In closing now, I offer this piece of advice "for what its worth" To all of you Mavic drone Pilots who are, or have experienced any annoying or inaccurate flight problems with your Mavics.

Just look at the idea of making your own simple, wooden or plastic rotating turntable and see if it improves landing accuracies, safety and the enjoyment factor for your drone flights.

If you do this, please share your findings here so we can all benefit!

Perfect example of anecdotal reports leading to false conclusions, flying in the face of all other known science. Just because event A preceded event B does not mean event A caused event B. Thousands of other independent intervening events could have been the real cause, assuming any cause even exists. Hence the need for a control group and double blind studies, and a statistically meaningful sample size, and statistically significant different results, which exceed mere chance. The results then need to be replicated by others around the globe, following the same scientific protocol. Start with @sar104 's suggestions above.

 

[Meta4]

In reply I will simply say again that the series of landing images I posted were not controlled manually, this was not possible before the calibrations made with the accuracy of my device.

Of all the improbable things you might think your compass and its calibration might affect, autolanding precision is about the least likely.
Your compass has nothing to do with how close your drone lands to the launch spot.

 

[gnirtS]

Compass calibration has absolutely nothing to do with precision RTH accuracy.
They're completely different systems.

One relies on satellite coverage and geometry, contrast and distinctive landing areas along with a specific method of taking off, the other.....doesnt.

 

[mereflyer]

OK You guys sound like you know what you are talking about.

I am only asking you all to think about the possibility that some of your opinions could be wrong!

Have you considered that the interactions between the various units of the of the IMU, GPS and compasses, handled so effectively by the software algorithms, can be sensitive enough to respond to this kind of calibrating accuracy, that normal rules as expounded by you good people may not be so clear cut!

No doubt each of you may have noticed, that many of our fellow Pilots are indeed having problems with their drones and from what I can see, there is very little help being offered by you experts.

So how about doing something more constructive - (like suggesting some effective fixes for their problems)
Make fixing these problems, more important than just criticizing my findings and results!

Whether you believe my findings or not, is of little consequence to me, I happily fly and enjoy a Mavic Pro that is definitely more accurate in flight and auto-landing on RTH than it was originally - I am convinced of the value of mechanically accurate Compass calibration, that is all I have to say on the matter - Thanking you all for your interest and contributions - Case closed!

 

[Meta4]

OK You guys sound like you know what you are talking about.
I am only asking you all to think about the possibility that some of your opinions could be wrong!

I've suggested something similar to you and backed it up with relevant information ... but that's been ignored.

Have you considered that the interactions between the various units of the of the IMU, GPS and compasses, handled so effectively by the software algorithms, can be sensitive enough to respond to this kind of calibrating accuracy, that normal rules as expounded by you good people may not be so clear cut!

I have ... and I'm certain that what I've put to you (which has been backed up by a very authoritative scientist) is accurate information.
Understanding these things helps much more than not understanding and saying, maybe this ... perhaps that ..

No doubt each of you may have noticed, that many of our fellow Pilots are indeed having problems with their drones and from what I can see, there is very little help being offered by you experts.
So how about doing something more constructive - (like suggesting some effective fixes for their problems)

1. I would disagree with your statement that many of our fellow Pilots are indeed having problems with their drones.
Your perception of issues is exaggerated and not supported by any evidence.
2. You haven't read much on the forum have you?

Make fixing these problems, more important than just criticizing my findings and results!

Again .. the problems you perceive aren't anywhere near as common as you think.
And you are talking to some of the flyers that do make an effort to help dispel myths, improve real understanding and solve mysterious flight incident issues.
Trying to clear up some of the misconceptions put forward in this thread is part of that effort.
I haven't seen any findings or results from you, just wild assumptions that aren't accurate or supported by facts.

Whether you believe my findings or not, is of little consequence to me ..
I am convinced of the value of mechanically accurate Compass calibration, that is all I have to say on the matter.

That's abundantly clear.

I happily fly and enjoy a Mavic Pro that is definitely more accurate in flight and auto-landing on RTH than it was originally

If you really believe that compass calibration affects your RTH landing accuracy, you have no understanding of how your drone works and should not be offering "advice".
That's like suggesting drivers service their car's air conditioning to fix transmission problems.

 

[Davros007]

Out of interest, are people calibrating them correctly? The words say to do it first held flat, but the second to do it ‘vertically’ - with a picture showing (DJI Go4) it held ‘sideways’. I assumed it was supposed to be held ‘camera up’ or down, and it worked. But then I also tried it with the drone ‘sideways’ as per the picture (which also worked) - An odd tendency to rise slowly in flight disappeared!

 

[DanMan32]

Put it this way:
First part of compass calibration is on the yaw axis with AC in normal flight position.
Second part of calibration is in roll axis with AC pointed up as if it were a rocket. Pointing down should also be OK.

Some do the rotations by spinning themselves with the AC held out in their arms. That's not the best approach.