Beta Testers requested to fix Compass Issue

[ianwood]

So, No news from DJI?
It´s a shame how long they are taking to resolve this serious issue. :shock:

Yes and no. We are all growing impatient and rightfully so. This bug should have never made it into production in the first place. However, once you have a bug in a production system, there's a lot of steps to putting a fix out into the wild.

They've been asking for more data from the "beta testers" and they are giving us updates on where they are with it. We need to give them some time to find the issue, fix it and issue a new firmware. Rest assured, we won't be letting them ignore or dismiss the issue. It will be fixed.

 

[lucianopacheco]

Thank you Ian.
We all have faith in you (not in DJI).

 

[dbot3000]

We all appreciate what you're doing Ian! DJI has had some missteps here and there but I have faith that they will address the issue. The fact that they are on boards like this and listening to users is a fantastic step that many companies never take. So that in itself is a great sign.

Tracking down, verifying, and fixing bugs like this does indeed take time. I'm happy to wait... (to a point, of course).

Best.

So, No news from DJI?
It´s a shame how long they are taking to resolve this serious issue. :shock:

Yes and no. We are all growing impatient and rightfully so. This bug should have never made it into production in the first place. However, once you have a bug in a production system, there's a lot of steps to putting a fix out into the wild.

They've been asking for more data from the "beta testers" and they are giving us updates on where they are with it. We need to give them some time to find the issue, fix it and issue a new firmware. Rest assured, we won't be letting them ignore or dismiss the issue. It will be fixed.

 

[Navman]

Scared to fly my new Phantom 2 after only 4 flights, constant battle to correct for counter clockwise TBE

On my 800 quad with Naza lite, setting the X Y and Z dialed it in never to be touched again.

Why don't we set the Z in this Naza?

My P2 has H3 3D gimbal, tall landing gear, AVL58 transmitter setup

Any advise appreciated as I'd like to try something before I head to the field and scare myself some more :shock:

 

[d4ddyo]

Scared to fly my new Phantom 2 after only 4 flights, constant battle to correct for counter clockwise TBE

On my 800 quad with Naza lite, setting the X Y and Z dialed it in never to be touched again.

Why don't we set the Z in this Naza?

My P2 has H3 3D gimbal, tall landing gear, AVL58 transmitter setup

Any advise appreciated as I'd like to try something before I head to the field and scare myself some more :shock:

You should be able to fly without a hitch in ATt mode at least enough time for it to autocorrect assuming your only problem is the gps/compass issue.

 

[Navman]

Scared to fly my new Phantom 2 after only 4 flights, constant battle to correct for counter clockwise TBE

You should be able to fly without a hitch in ATt mode at least enough time for it to autocorrect assuming your only problem is the gps/compass issue.

Thank you may have to give that a try and see if "learns" once warmed up

Planned on filming a friends 2 million dollar home on the river tommorrow, one of the reasons for buying this machine.

Really impressed with the liquid smooth video though :mrgreen:

 

[ianwood]

On my 800 quad with Naza lite, setting the X Y and Z dialed it in never to be touched again.

Why don't we set the Z in this Naza?

I don't know the regular Naza set up. Can you elaborate on what you described?

Thank you may have to give that a try and see if "learns" once warmed up

It learns but it also forgets. As in it takes 8 minutes in Los Angeles for the Phantom to learn to fly straight. Once you shut it off, the learning is wiped out and will take another 8 minutes to fly straight on the next battery. Fun!

 

[d4ddyo]

Scared to fly my new Phantom 2 after only 4 flights, constant battle to correct for counter clockwise TBE

You should be able to fly without a hitch in ATt mode at least enough time for it to autocorrect assuming your only problem is the gps/compass issue.

Thank you may have to give that a try and see if "learns" once warmed up

Planned on filming a friends 2 million dollar home on the river tommorrow, one of the reasons for buying this machine.

Really impressed with the liquid smooth video though :mrgreen:

Actually you should try filming in ATT mode. Just know you will get drift but use the drift to your advantage.

 

[Navman]

Ian, on the older Naza puck we had to tell the program where the puck was located relative to center of the machine forward/back, side to side and Z the up and down center of gravity of the machine.

Got the first two correct on the 800, but guessed way off on Z, man it was a work out to hold in one place for framing a photo.

Once I actually measured the cm, it would just sit there in the sky asking me "Now what do you want me to do next?"

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done!

Thank you for all you and the others are doing to get this figured out and I'm sure DJI is working hard on it too.

On my 800 quad with Naza lite, setting the X Y and Z dialed it in never to be touched again.

Why don't we set the Z in this Naza?

I don't know the regular Naza set up. Can you elaborate on what you described?

Thank you may have to give that a try and see if "learns" once warmed up

It learns but it also forgets. As in it takes 8 minutes in Los Angeles for the Phantom to learn to fly straight. Once you shut it off, the learning is wiped out and will take another 8 minutes to fly straight on the next battery. Fun!

 

[Navman]

Actually you should try filming in ATT mode. Just know you will get drift but use the drift to your advantage.

Hmmm... never thought of it that way, will have to give it a try.

Spent better part of the day upgrading to 3.02 to get the slow gimbal and mount LED tail light etc., now it's to hot outside

 

[nzvideoguys]

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done![/quote]

Exactly! The solution coming from DJI back then was to rotate the compass. Sadly, very few of the beta testers here seem to be aware of this as they've probably never flown anything but a Phantom, (I could be wrong). I also bought up the point about dialling in the X Y and Z distances, as these values are defaulted at "0" on the Phantom software, which, when accurately measured, is not their true position. The GPS on the Phantom is located in the centre, under the plastic body shell, and the compass is way down on the right rear leg. If we had a version of the software where the XYZ values could be changed to reflect their actual position, we may be a step closer to solving this complicated riddle.

 

[xtonex]

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done!

Exactly! The solution coming from DJI back then was to rotate the compass. Sadly, very few of the beta testers here seem to be aware of this as they've probably never flown anything but a Phantom, (I could be wrong). I also bought up the point about dialling in the X Y and Z distances, as these values are defaulted at "0" on the Phantom software, which, when accurately measured, is not their true position. The GPS on the Phantom is located in the centre, under the plastic body shell, and the compass is way down on the right rear leg. If we had a version of the software where the XYZ values could be changed to reflect their actual position, we may be a step closer to solving this complicated riddle.[/quote]


Do we know why they decided to not allow axs to the settings on the phantom naza's?

 

[sar104]

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done!

Exactly! The solution coming from DJI back then was to rotate the compass. Sadly, very few of the beta testers here seem to be aware of this as they've probably never flown anything but a Phantom, (I could be wrong). I also bought up the point about dialling in the X Y and Z distances, as these values are defaulted at "0" on the Phantom software, which, when accurately measured, is not their true position. The GPS on the Phantom is located in the centre, under the plastic body shell, and the compass is way down on the right rear leg. If we had a version of the software where the XYZ values could be changed to reflect their actual position, we may be a step closer to solving this complicated riddle.

But the Phantom approximates to a rigid body, so only the orientation of the compass matters, not its location. So if you could set the X, Y, Z values, you would choose the GPS module location, but presumably that is hard-wired into the programming since the GPS module is always in the same position in the Phantom.

 

[nzvideoguys]

But the Phantom approximates to a rigid body, so only the orientation of the compass matters, not its location. So if you could set the X, Y, Z values, you would choose the GPS module location, but presumably that is hard-wired into the programming since the GPS module is always in the same position in the Phantom.[/quote]

I know there were separate areas to fill in the XYZ values for both IMU and GPS. The puck contained the GPS and the compass and DJI's remedy was to rotate the compass, in the case of NZ, it was rotated to the 2 o'clock position, or roughly 60 degrees.

 

[sar104]

But the Phantom approximates to a rigid body, so only the orientation of the compass matters, not its location. So if you could set the X, Y, Z values, you would choose the GPS module location, but presumably that is hard-wired into the programming since the GPS module is always in the same position in the Phantom.

I know there were separate areas to fill in the XYZ values for both IMU and GPS. The puck contained the GPS and the compass and DJI's remedy was to rotate the compass, in the case of NZ, it was rotated to the 2 o'clock position, or roughly 60 degrees.

Right - it cares about the IMU location and orientation, the GPS location (but not orientation), and the compass orientation (but not location).

 

[ianwood]

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done!

Exactly! The solution coming from DJI back then was to rotate the compass. Sadly, very few of the beta testers here seem to be aware of this as they've probably never flown anything but a Phantom, (I could be wrong). I also bought up the point about dialling in the X Y and Z distances, as these values are defaulted at "0" on the Phantom software, which, when accurately measured, is not their true position. The GPS on the Phantom is located in the centre, under the plastic body shell, and the compass is way down on the right rear leg. If we had a version of the software where the XYZ values could be changed to reflect their actual position, we may be a step closer to solving this complicated riddle.

Most of us are aware that other Naza platforms rotate the compass. DJI advises not to rotate the compass on the P2. Rotating any compass to an off axis position, P2 or not, is generally a bad idea unless the software compensates for it. When you rotate the compass off axis, you have at least two axes that no longer have the correct sensitivity because they aren't aligned with the orientation of the vehicle. As result, you wind up with inaccuracies. When you apply a tilt compensation algorithm to the compass heading it gets worse as the compass is not moving through the same axes as the vehicle. I would suspect a number of very expensive fly aways on DJI's higher end platforms may have been caused by this.

The right way to correct for declination is painfully simple: You have a lookup table of declination data which is roughly 3Kb to 4Kb in size and based on GPS lat/lon. Once the GPS position of the vehicle is known, the corresponding declination is pulled from the lookup table and applied to the magnetic heading. It's about 3 lines of code and a 2 axis table. Pretty simple. My suspicion is that the P2 Naza is actually doing this but it is applying it backwards (i.e. subtracting as opposed to adding) which is why many have to rotate their compass by roughly twice their declination.

Anyway, rotating the compass is not very smart on any platform. At best, it's a work around to a problem that shouldn't exist.

 

[d4ddyo]

This declination shouldn't come as a surprise to DJI, we all went thru it with the combo mag/compass thingy, rotate puck according to your magnetic declination, Done!

Exactly! The solution coming from DJI back then was to rotate the compass. Sadly, very few of the beta testers here seem to be aware of this as they've probably never flown anything but a Phantom, (I could be wrong). I also bought up the point about dialling in the X Y and Z distances, as these values are defaulted at "0" on the Phantom software, which, when accurately measured, is not their true position. The GPS on the Phantom is located in the centre, under the plastic body shell, and the compass is way down on the right rear leg. If we had a version of the software where the XYZ values could be changed to reflect their actual position, we may be a step closer to solving this complicated riddle.

Most of us are aware that other Naza platforms rotate the compass. DJI advises not to rotate the compass on the P2. Rotating any compass to an off axis position, P2 or not, is generally a bad idea unless the software compensates for it. When you rotate the compass off axis, you have at least two axes that no longer have the correct sensitivity because they aren't aligned with the orientation of the vehicle. As result, you wind up with inaccuracies. When you apply a tilt compensation algorithm to the compass heading it gets worse as the compass is not moving through the same axes as the vehicle. I would suspect a number of very expensive fly aways on DJI's higher end platforms may have been caused by this.

The right way to correct for declination is painfully simple: You have a lookup table of declination data which is roughly 3Kb to 4Kb in size and based on GPS lat/lon. Once the GPS position of the vehicle is known, the corresponding declination is pulled from the lookup table and applied to the magnetic heading. It's about 3 lines of code and a 2 axis table. Pretty simple. My suspicion is that the P2 Naza is actually doing this but it is applying it backwards (i.e. subtracting as opposed to adding) which is why many have to rotate their compass by roughly twice their declination.

Anyway, rotating the compass is not very smart on any platform. At best, it's a work around to a problem that shouldn't exist.

Your explanation makes total sense ianwood. Get them to fix it. LOL. I keep forgetting about the J hook and I nearly J-hooked into a tree today.

 

[ianwood]

I think this sums it up nicely. DJI, please pay attention to this:

 

[xtonex]

Lets all move to chicago! lol

 

[Xrover]

Ian,
You should be head of engineering for DJI.