phantom suddenly possessed in GPS mode

[MILLER4PRESIDENT2020]

the only thing that has changed in the course of a few flights was i bought the carbon fiber folding landing gear kit and installed it. I am going to go back and try the stock plastic ones but it is dark out so I wont be able to see if that fixes it until tomorrow so I figured i'd get a headstart on here.

so anyways everything was working gravy, bought and installed foldable carbon fiber landing gear from helipal
http://www.helipal.com/carbon-fiber...hantom.html?osCsid=02u0ng3unu6hnakqcvlvme9d83
installation was relatively easy except the compass doesnt fit around the new landing gear, so I had to sand down a small section of one leg in order to get the compass to be able to fit around and then tighten the screws.

take it out this morning to go flying, lifts off straight and hovers for a second, as soon as I go forward in GPS mode it starts drifting like crazy, I am fighting it and the controls seem all confused. I switch it to ATT before it decides to fly into a tree and bring it back and rescue it. This is a new flying area and there is one power line a couple hundered feet away so i thought maybe that is causing a problem. I took the phantom to another new area with no power lines, run the GPS calibration, lifts off fine in GPS and then as soon as i give forward axis it takes off. Put it back into ATT and bring it back to me and try GPS once more and you can see the thing visibly tilt as if its a killer drone from terminator 2 who just spotted his target and rushes off, as soon as i switch back into GPS. Enough drama for one session. Keep in mind both of these times the GPS acquisition process went flawlessly.

I bring it back home and plug it into the computer and run the IMU calibration multiple times, advanced and basic. No problems. I even referenced this video which I have never seen before.
[youtube]http://www.youtube.com/watch?v=io9bBh5_eyo[/youtube].

I start thinking to myself made I caused the compass to go out of whack when unscrewing and rescrewing the 4 screws. While i have it plugged in i start moving my screwdriver around and sure enough it causes the values to fluctuate. Colin says the compass raw values should be in between negative 300 and positive 300 in the video. My X and Y were fine but the Z was something like 395. MOD is supposed to be around 1600, which mine was. So one of my values is slightly off but keep in mind I have received no signals that my GPS cannot align during the warmup phase. I decided to mount the compass right on top of the phantom thinking maybe, just maybe the compass touching carbon fiber instead of plastic is throwing it off even though this is not likely So i take it out into my backyard, this time on a fishing line in case it decides to return to china. Go thru the satellite acquisition just fine, even calibrate the compass one more time to make sure. Take off and everything appears to finally be working, crisis averted...or so i thought.

I then drive 5 minutes down the road to the soccer fields, a usual spot ive flown tons of times before. Plug in with a brand new battery, charged but never used before. Don't bother with a calibration since i'm only 5 minutes from my last calibration, go thru the warmup sequence and all lights are OK. Take off in GPS and immediately non stop flashing red, to the point where it wants to go into autoland. Hmmmm, pull out the battery tester and it says 12.1 volts. Maybe its just a fluke battery out of the factory? I switched to the factory battery which has never had a problem, test the voltage...OK, plug it in fire up the phantom, GPS satellite acquisition no problem, take off, hover, move forward, and then it is suddenly possessed again!!!!! I'm talking the controls are almost fighting something in GPS mode, it has it's mind made up that it wants to go some direction and full power will not stop it, adding power and using the controls does not correlate properly to the movements that should be occuring. So now I am at home, going to put back on the old plastic legs but this is driving me nuts, and quite frankly I'm scared to fly it until I get some reassurance.

The only other variable in this whole equation is that I used electrical tape now to tape off the transmitter and compass, when it came straight out of the box they gave you some of that white tape that resembled medical tape. It's not like i wrapped 15 feet of electrical tape around the transmitter and compass sealing either of them off from the outside world, maybe 2 inches, just enough to wrap around the leg twice.

Help, ive gone from awesome to possessed in the course of a day.

 

[OI Photography]

If you have your compass mounted the way it is in those product pics at helipal, then that's your problem. It must be mounted facing inboard (towards the other rear leg), and perpendicular to the ground. I learned the same lesson when I first tried an aftermarket CF landing gear (not the same as yours though).

 

[MILLER4PRESIDENT2020]

it just may be that simple, i seem to remember something about the direction the compass needs to be mounted when originally building my phantom but have long forgotten that. but just to clarify with the picture here what i need to do. (yes i stole the pic off of helipal and photoshopped it) your saying i need the flat side of the compass to be facing number 3? B y the looks of the old plastic landing gear you're also saying it is supposed to be perfectly vertical up and down? this new landing gear is at a slight angle. hmmm, guess im going to have to fashion parts which totally defeats the original weight saving purpose

 

[tanasit]

http://www.shapeways.com/model/1075118/ ... erialId=62

http://www.shapeways.com/model/1019077/ ... erialId=62

http://www.shapeways.com/model/1396337/ ... terialId=6

I concur that the compass has to be installed perfectly vertical and that's why several after market landing gears provide the "off-set" compass mount as seen above.
In addition, the relative vertical distance of the compass from the NAZA has to be the same as the stock setup.

You can even test if the GPS and compass if they work correctly by performing the IOC function because:
CL requires compass and the HL requires GPS to work correctly.
The flight path should be as seen below:

 

[MILLER4PRESIDENT2020]

update, the problem is fixed. that **** compass is supersensitive i guess. hopefully my stupidity can help someone else avoid this problem.

 

[tanasit]

update, the problem is fixed. that **** compass is supersensitive i guess. hopefully my stupidity can help someone else avoid this problem.

Could you explain in detail what went wrong so that we all can learn?
Or just because the compass is NOT vertical?
Thanks,

 

[LeoS]

Did the tape skewed the compass's angle somehow?

 

[MILLER4PRESIDENT2020]

if you look at the carbon fiber legs in the pictures, they are not 180 degrees up and down. probably like 20 degrees off vertical, that was enough to make it out of whack, and so then i got the bright idea to mount the compass on top of the phantom thinking it would get the best sattelite signal possible....wroonggggggg. that made it even worse. It HAS to be perfectly up and down and facing the number 3 leg like on your original landing gear. I fashioned a temporary mount out of tape that works now, i will come up with a more permanent looking one soon.

 

[tanasit]

if you look at the carbon fiber legs in the pictures, they are not 180 degrees up and down. probably like 20 degrees off vertical, that was enough to make it out of whack, and so then i got the bright idea to mount the compass on top of the phantom thinking it would get the best sattelite signal possible....wroonggggggg. that made it even worse. It HAS to be perfectly up and down and facing the number 3 leg like on your original landing gear. I fashioned a temporary mount out of tape that works now, i will come up with a more permanent looking one soon.

The compass does NOT get the signal from the satellites but from earth magnetic field!
The GPS receiver that was mounted under the dome is responsible for satellite signal detection.

 

[Dave Pitman]

In addition, the relative vertical distance of the compass from the NAZA has to be the same as the stock setup.

Can you elaborate on this statement please. I don't understand how the vertical offset can make a difference. The standard naza package integrates the compass into the gps puck, and is certainly at a different offset to the fc in contrast to the Phantom.

 

[OI Photography]

In addition, the relative vertical distance of the compass from the NAZA has to be the same as the stock setup.

Can you elaborate on this statement please. I don't understand how the vertical offset can make a difference. The standard naza package integrates the compass into the gps puck, and is certainly at a different offset to the fc in contrast to the Phantom.

You know how the NAZA assistant has a place you can edit the offset of the GPS location (which is always the same in the Phantom anyway)...it's the same idea for the compass. The NAZA assumes that the compass is exactly so far below and so far to the side of the center of the NAZA, and adjusts it's calculations accordingly. You just can't adjust the values in the software for the compass, so it needs to match the stock location to work with the hard-coded numbers. In the NAZA v2, adjustments to the "GPS offset" value would also affect the compass calculations.

 

[Dave Pitman]

You know how the NAZA assistant has a place you can edit the offset of the GPS location (which is always the same in the Phantom anyway)...it's the same idea for the compass. The NAZA assumes that the compass is exactly so far below and so far to the side of the center of the NAZA, and adjusts it's calculations accordingly. You just can't adjust the values in the software for the compass, so it needs to match the stock location to work with the hard-coded numbers. In the NAZA v2, adjustments to the "GPS offset" value would also affect the compass calculations.

So, you are saying that the Naza M V2 Assistant software is different than the Assistant software we are using which goes by the same name?

In other words, how does the assistant software know my Naza M V2 is in a Phantom, and not a Flamewheel? Or, that I have not changed out my current gps and compass for a puck?

Sorry if I'm missing something simple.

 

[GeneL]

In addition, the relative vertical distance of the compass from the NAZA has to be the same as the stock setup.

Can you elaborate on this statement please. I don't understand how the vertical offset can make a difference. The standard naza package integrates the compass into the gps puck, and is certainly at a different offset to the fc in contrast to the Phantom.

This is just conjecture, OK, since I think it's fair to say that only a relative handful of people really know and understand what kinds of programming voodoo really goes on inside the Naza. However, the compass is wired in two completely configurations between the Phantom, with the compass on the leg, and the Flamewheel type with the compass and GPS integrated in a puck. It's not unreasonable to presume that the Naza can recognize the difference and respond accordingly. Still, similar to the leg mount, the puck compass is expected to be mounted in a very specific orientation or problems will occur.

Moreover, I suspect it's a mistake to entirely separate the functions of the compass and the GPS with regard to the ultimate behavior of the aircraft, because I think that the Naza algorithms are synthesizing both of them together to stabilize and control the multi in the mode(s) that they are both supposed to be engaged, so that it's hard to determine which one is causing abnormal flight. So, when we're flying in "GPS" mode, the GPS isn't just functioning to know absolute position, and the compass isn't only responsible for moving in a given direction. I think that input from all the sensors (including gyro,accelerometer, etc.) figure in to the final control output, or the part we see as the way the aircraft flies.

So, I guess what I'm trying to say that mounting the compass In a non-standard position or orientation isn't going to effect just what direction the aircraft is told to go, but also could result in a disruption of the entire stabilization system. Likewise, a compromised GPS won't only cause the multi to lose position, but could cause other behavior problems that would seem to be unrelated the aircraft location alone.

Again, these are just my suspicions,not my convictions, but I would sure like to know how far off the mark they are and why.

 

[GeneL]

So, you are saying that the Naza M V2 Assistant software is different than the Assistant software we are using which goes by the same name?

In other words, how does the assistant software know my Naza M V2 is in a Phantom, and not a Flamewheel? Or, that I have not changed out my current gps and compass for a puck?

Sorry if I'm missing something simple.

The Assistant has a tab for entering the position of the GPS when used for the Phantom as well as the Flamewheel. It's just that the Naza and the GPS are hard mounted at the CG (before additional gear is added) on the Phantom, so the entries are all zero. On the Flamewheels, the puck is commonly mounted wherever the builder cares to put the pylon, so the xyz coordinates have to reflect the offset from the CG, wherever that ends up. The Assistant is the same for both, however.

 

[tanasit]

Re: another possible improvement

http://www.youtube.com/watch?v=f6oDVQwUPwU

http://www.youtube.com/watch?v=cYUfQp5caHY

http://www.multirotorforums.com/showthr ... eclination

When one decides to modify or build the new quad, he will run into some troubles some of which may not be obvious, in particular when we are dealing with software and setting parameters are the main key for success.
Relocation of the compass in Phantom may not seem obvious especially when there are more than one modifications involved. But through the process of one by one elimination, it was discovered that not only the compass has to be perfectly vertical but the relative vertical distance with respect to the NAZA has to be the same as the stock setup. Technical reasons behind this have yet to be fully explained.
More over, a little less known for GPS receiver installation is the "Magnetic Inclination Compensation" (see links above). Think of it as a fine tune for your GPS system, depending on your location you may want to adjust your GPS accordingly.

 

[Dave Pitman]

Guys,

To be clear, I have no doubt that the orientation of the compass is critical. However, I can find no documentation of any set vertical offset specification whatsoever. Remember, the compass is telling the Naza which way is magnetic North relative to the front of the Model, nothing more.

I don't have time to watch the videos posted right this minute. Are they supposed to document the need for precise vertical placement of the compass?

The Naza flight control system is indeed complex and picky. But, we don't need or want to make it more complicated than it is. If vertical offset (not orientation) was critical, I suspect there would be a spot in the Assistant to specify it, but I'm willing to be corrected with proof.

 

[Ton4]

So, has the Naza a map of the world with all inclinations? Or does it just assumes the magnetic north is the true north in its calculations for RTH? (so it will not fly in a straight line home, and constantly corrects course). I read somewhere pilots in Greenland had some problems (high declination) at the other hand there are nice Phantom video's from that area. Someone closer to the North Pole was unable to fly gps, made the Phantom go bezerk. Would "funny" magnetic fields confuse or crash the Naza? Cause fly-aways?

Sent from my SM-T311 using Tapatalk

 

[GeneL]

Someone closer to the North Pole was unable to fly gps, made the Phantom go bezerk. Would "funny" magnetic fields confuse or crash the Naza? Cause fly-aways?

Sent from my SM-T311 using Tapatalk

The DJI docs specifically state that compass calibration dos not work in the polar circle.

Also, there is a place for a vertical offset in the assistant, right alongside the entries for x and y offset. It is the "z" offset. Again, these settings are apparently hard wired for the Phantom, because the Phantom's compass and Naza are supposed to be at a fixed, preset location. On a Flamewhel, though, both are located according to the builders preference; hence, the need to supply numbers for the offset of the GPS/Compass puck from the CG.

 

[Ton4]

If the gps and / or compass would be on 2 meters distance of the cg or 2cm or 20cm should not matter for navigation purposes. So why would this distance be so important?

Sent from my SM-T311 using Tapatalk

 

[Dave Pitman]

\Also, there is a place for a vertical offset in the assistant, right alongside the entries for x and y offset. It is the "z" offset. Again, these settings are apparently hard wired for the Phantom, because the Phantom's compass and Naza are supposed to be at a fixed, preset location. On a Flamewhel, though, both are located according to the builders preference; hence, the need to supply numbers for the offset of the GPS/Compass puck from the CG.

I see the offsets (X,Y,Z) for the GPS. Please provide documentation for Compass offset

(For others that "believe" the vertical compass offset is critical, please do not respond without documentation of some sort, thanks)