Welcome to PhantomPilots.com

Sign up for a weekly email of the latest drone news & information

Toliet Bowl GPS hover fix, fixed mine, now rock steady

Discussion in 'Phantom 1 Help' started by denodan, Sep 6, 2013.

  1. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    Anyone having trouble with drifting in GPS, or toilet bowl effect? Well out of the box my was bad, and looking around the internet, it seems there is a simple fix.

    What you need to know is your magnetic declination:

    http://magnetic-declination.com/

    When you have found out how far east or west you need to go, then loosen your compass, and turn it around so it will match the magnetic declination angle.

    I had to turn mine clockwise, so it is between 12 and 1 o clock position. If you see your Phantom doing a clockwise circle, then you got to move your compass clockwise, if anticlockwise, move anti clock wise. I adjusted mine several times and now is almost rock steady. Apart from a stop, will drift slightly counter clockwise for a couple of seconds then hover rock steady.

    You can fine tune by moving your compass around on the leg. Think I got mine pretty good now.
     
  2. marcus_canada

    Joined:
    Jul 22, 2013
    Messages:
    304
    Likes Received:
    4
    The fix for the toilet-bowl-of-doom issue I picked up after installing 3.16 was to run through the advanced IMU Calibration and do the Compass dance.
    It was rock steady on 3.12 and remains rock steady in 3.16 after the cal.
     
  3. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    You got the old Phantom? adjusting your compass will fix it. Think the old Phantoms are not fully compatibly with 3.16, I think 3.16+ is really for the newer Phantoms.

    Think most of the problems are happening with people updating the older Phantoms to 3.16, more so then the newer ones? I could be wrong? But have a feeling this maybe the case.
     
  4. marcus_canada

    Joined:
    Jul 22, 2013
    Messages:
    304
    Likes Received:
    4
    I don't know is the simple answer, it came with 3.12, I didn't upgrade it to that. It was my understanding that 3.16 was for [but not restricted to] the updated board and DJI Gimbal.
     
  5. martcerv

    Joined:
    May 27, 2013
    Messages:
    1,002
    Likes Received:
    2
    Running 3.16 on an older phantom with no issues here, did load defaults and properly calibrated everything and its done lots of flights since. I think the main issue may be people updating and then doing advanced calibration, the window says the NAZA is too hot or something along those lines, to turn it of for 10 minutes then redo the advanced calibration.

    I have seen many people say to just keep clicking through the warning and then not doing as it says flying after a bad calibrations. This could be a major issue with the 3.16 update and if you just click through a warning that pops up 5 or 6 times with clear info on how to do this properly without ever reading the warning, you cant be surprised too much if things go bad after.
     
  6. Fdnyfish

    Joined:
    Jun 6, 2013
    Messages:
    664
    Likes Received:
    36
    Location:
    Barnesville, PA
    Where are you getting this information that 3.16 is for only the newer Phantoms.

    And what is a newer phantom? The Vision? That has not been released yet.

    3.16 is compatible with ALL Phantoms.

    http://www.dji.com/whats-difference-of- ... naza-m-v2/


    On a side note..... 3.19 is only compatible with the gold plated phantoms available in the UK only
     
  7. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    The newer Phantoms have the 2 black counter clockwise nuts and the newer Naza M V2, and maybe other slight board mods to the hardware? The older Phantom has the Naza M, so there maybe a ever so slight difference in hardware to cause problems. It happens in PC's all the time, run the same software and operating system, and the ever so slight hardware can cause slight problems.

    It is just a thought, not fact, as it does seem those with the older Phantoms maybe having the biggest problems with 3.16 update, so maybe due to the slight hardware problems, in the fact 3.16 maybe more compatible with the Naza M V2 then the older Naza controller. Also another slight difference is the newer Phantom will fly longer on the same standard battery.

    Sometimes ever so slight hardware changes can cause problems, not always, but can happen.
     
  8. Wait!comebak

    Joined:
    Aug 12, 2013
    Messages:
    14
    Likes Received:
    0
    When you have found out how far east or west you need to go, then loosen your compass, and turn it around so it will match the magnetic declination angle.

    Denodan, The compass looks as it is shaped to fit on the leg only one way? How did you overcome this?
    Cheers
     
  9. bumper

    Joined:
    Aug 6, 2013
    Messages:
    94
    Likes Received:
    0

    Is this fix from DJI? Or is it an experiment posted on a forum? (Not saying such research isn't often accurate.)

    Reason I ask is the fix doesn't make sense to me*. When you calibrate the compass, as indicated by the green light going out, you have just adjusted for the local conditions including inclination and declination. If the compass won't calibrate, you can use the "magnet trick", as shown on one of DJI's videos, to "degauss" the flux sensor.

    *Consider for a moment that the Phantom, when hovering, only knows what direction it is pointing based on the compass. So if rotating the compass X number of degrees, that only alters what direction the Phantom thinks it's pointing by that same number of degrees. That would have no effect on toilet bowl effect. Toilet bowl effect is doubtless caused by the normal GPS dithering inaccuracy, which is typically on the order of several meters.

    bumper
     
  10. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23

    All I know it works well, the compass calibrates to magnetic North, not true North. GPS needs true North to be accurate, and this is why you got to turn your Compass to off set it, so it is calibrated to True North. If you calibrate to Magnetic North, the GPS sees this as an error, as GPS will work on True North, so if your compass is set to Magnetic North, and GPS is working on True North, then you got a error straight away, this is what causes the circular motion, as you have a compass error, and GPS cannot fully lock on it due to the error it sees in your compass, so like getting two conflicting readings of the compass.

    All I know it works, and works and cures toilet bowl effect.

    Also you can loosen your compass far enough to turn it one way or the other and re tighten it. I did read it on here somewhere, were one or 2 other people tried this fix and works.

    So if you got this problem offset your compass to your magnetic declination offset, the nearer you are the more rock steady your Phantom will be, no wondering, or drifting, but hold steady, mine holds rock solid now, may have a slight drift when first lifting off for a few seconds, but locks in solid. I had bad toilet bowl out of the box, and this cured it for good.

    The screws on the compass are quite long and enable you to loosen it enough to move and re tighten when you got it right.
     
  11. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    So when your calibrating your compass, it is to magnetic North, off setting the compass to your magnetic declination, your calibrating your compass to true North, but don't think you have to be dead accurate. What I did was loosen the compass turned it took off looked for drift or circular, or rectangular motion, so adjusted it till I go no motion, or hardly any, when I was happy I got no drift, or the least drift, then tightened the compass into the off set position, and as a result for a few seconds on takeoff, a very slow drift, then locks solid or as near as possible.

    You may not get it perfect, as each Phantom will very slightly, but will either get no drift, or hardly any. But give it a few secs to see what happens, it may lock in solid, or only slight drift, but you will improve it big time.
     
  12. bumper

    Joined:
    Aug 6, 2013
    Messages:
    94
    Likes Received:
    0
    GPS is autonomous, it needs no external magnetic compass input to determine its direction or orientation (the IMU does need the magnetic compass input to determine orientation when not moving as in a hover). GPS locates lat and lon position as well as "GPS altitude" based solely on sat triangulation. When the GPS moves, it can then determine course direction, speed etc, based on it's changing position. Though even then it doesn't know which direction it's pointing, if you are facing north, but going in reverse, the GPS sees this as going south and is okay with that (see glider example below).

    I do not have a clue how NAZA M's algorithms work, or for that matter how the mag compass input input is actually utilized. But for the hover to be accurate, it seems that even with the flux sensor rotated to a different position on the leg, it should make no difference, assuming compass cal is current, as the NAZA should only need to know about the Phantom's physical rotation, if any, in order to keep it pointed consistently. The GPS doesn't give a hoot about it's rotational position - - only about movement to calculate course direction.

    As an aside . . . I was flying my glider in mountain wave at just under 18K' into a 70 knot headwind. As I slowed my speed below 70, the glider is first stationary over the ground and then moving backwards slowly. The GPS moving map is set to show "track up" (meaning the display is automatically positioned with the glider's heading at the top of the screen. As the glider moves backwards, the moving map display flips 180 degrees so even though pointing west, the map shows my pointing east. GPS, without movement or some other input (magnetic in DJI's case) has no clue. It IS weird having the map show backwards!

    bumper
     
  13. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    Still it works, rotating the compass on the leg to magnetic declination, or near works, so Naza, or something else in the Phantom must look at true north and checking your compass as well as GPS compass, so since Magnetic North and True North are not the same you would get a compass error, which is the toilet bowl effect. Whatever it is, I know it works, Also in the Phantom Owners facbook page Richard has his set to the 1 o'clock position and works.
     
  14. bumper

    Joined:
    Aug 6, 2013
    Messages:
    94
    Likes Received:
    0
    denodan,

    Not meaning to be argumentative.

    If it works for you, and I'm certainly not disputing that, then its working by PFM (pure f-ing magic).

    Again, in a hover, the GPS is only providing lat and lon positioning to withing several meters at best, and GPS altitude, due to the geometry of it all is far less accurate, on the order of + or - 100 feet or more. Compass error and direction it's pointing should be irrelevant as long as the compass input is giving the Phantom a relatively consistent circle of 360 degrees - whether that's relative to true north or 20 degrees off on way or another shouldn't matter - - it's all relative as far as the data going to the IMU is concerned.

    Magnetic compasses can almost never be calibrated to accurately reflect even magnetic direction on all points of the compass. That's why ships and airplanes have compass correction cards, to record the error on each cardinal point. Save goes with the flux gate compass.

    This'll be my last post on this topic though . . . I'm clearly not getting the point across, sorry.

    bumper
     
  15. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23
    Well put it this way, you don't have to get it, it works, simple as that, and the fix is linked with off setting your compass to magnetic declination, don't know how it works, you don't have to get it, all I know is it works?

    Ask Richard in the Facebook Phantom owners group, he has his compass offset to the 1 o clock position and works. maybe someone can explain how it works? All I know is it does.
     
  16. marcus_canada

    Joined:
    Jul 22, 2013
    Messages:
    304
    Likes Received:
    4
    Also not trying to be argumentative but the compass is getting a bad rep here and there is a lot of misunderstanding surrounding it.
    The first thing is it's name, compass. I have issues with that straight off.

    It is a fixed point in space which the Nazza uses to reference in relation to the GPS position for chassis orientation purposes only.
    As Bumper says, the phantom could be located anywhere on the planet and the Magnetic North direction could be different from the factory position's North bearing by quite a way so if the compass on the leg was truly depending on the magnetic mapping it would only work correctly in locations where the magnetic North direction was the same as the factories. The actual position of Mag' North is irrelevant to the Phantom for orientation purposes, and further, totally irrelevant to any other of it's functions.

    Navigation, in this case a term limited to "return to Home", has nothing to do with N/S/E/W, It's all done on absolute Lat/Long.

    The 'Compass' is only fully exploited for IOC function where the crafts Yaw position is particularly important.

    The IMU's are the things responsible for simple Pitch/Roll/Yaw stability.

    If your bird is unstable in Yaw, you need to do the compass dance. If it's unstable in pitch/roll you need to do the simple or advanced IMU calibration.

    A tiny movement on the position of the compass isn't going to do much but don't move it a long way, like onto the wrong leg, because this fixed position of it in relation to the GPS 'centre' is fixed in firmware and cannot be reset or adjusted.
    Moving the Compass and doing 115 compass dances won't do Jack for you.

    If you don't believe me ask Jimmy, my neighbor. He doesn't know any more than me but if we are offering references............. :D
     
  17. denodan

    Joined:
    Feb 8, 2013
    Messages:
    359
    Likes Received:
    23

    Well have you tried it? It does work and cured my Phantom. The compass dance fixed stuff all, this fix works and my Phantom has never held as steady as it does, till I moved the compass around. It is still on the same leg, just turned it. Seems your arguing it does not work, but I can tell you it certainly does work, and MUI calibration also did stuff all, the only fix that worked was turning the compass.
     
  18. MrMediaGuy

    Joined:
    Jul 31, 2013
    Messages:
    171
    Likes Received:
    1
    Very old thread I realize, but this problem came up today on another thread and I thought I'd chime in here in case anyone finds this debate later on. :)

    All the information from Bumper and others on GPS here is correct. However, the thing we don't know is *exactly* what the NAZA does with the combination of compass and GPS data. (It may in fact be doing something with the compass data that is not strictly correct in terms of relating it to GPS data and then to control outputs.) Regardless, adjusting the orientation of the compass is the accepted means of fixing the TBE problem. Think about the NAZA M-V2 on F450/550 frames, where the compass is built into the GPS puck. The puck must be installed exactly parallel to the main axis of the frame -- and if your magnetic deviation from true north is very high (extreme latitudes) then you do have to turn the puck by the appropriate number of degrees to compensate.

    I can only speculate, but we know that when the quad is moving it has a sense of direction of the frame coming from GPS -- i.e., when there is movement the NAZA "knows" from the NMEA data sentences in which direction that movement is happening, but it does not know in which direction the frame is pointing. And without knowing that, it can't compensate with the correct motor inputs.

    Enter the compass, which tells the NAZA, "GPS says you're moving south, but you're pointing east, so to maintain position you need to adjust motors 2 and 4." (Just an example.) On the other hand if you were moving south but pointing north, the main control signals would be to the *other* two motors. (And of course in reality the control affects all 4 in some way, but I'm simplfying for the purpose of explanation.)

    The issue is that GPS uses true north, while the compass uses magnetic north. At normal latitudes the deviation is minimal. But at very extreme latitudes the deviation is a number of degrees, and that makes a huge difference.

    Of course, because the GPS knows your latitude, the NAZA could easily take the compass data and calculate true north from magnetic north and solve the problem with math rather than making you rotate the compass or puck. And in fact they did exactly that way back in March 2013. So I guess the whole discussion is academic unless you are running really old firmware. :)
     
  19. Buk

    Buk

    Joined:
    Feb 28, 2013
    Messages:
    654
    Likes Received:
    29
    Location:
    Iowa, United States
    Warning - a noob about to make a comment:

    Why would the GPS of the Phantom care about true or magnetic north? Doesn't it just care about two positions, Home and Current as determined by six or more satellites.

    Home for RTH function and IOC. Current for hovering. What would north have to do with either of these locations - just lat and long. Yes, I know it can be related to north. You only need to know two points, where you came from and where you are.

    Compass just provides a point of reference to determine the front of the Phantom. So that when operating in non-IOC or non-RTH control, pitch and roll relative to the front of the Phantom. The compass dance helps determines where that leg on the landing gear is located and as such the front of the Phantom. My question would be, does rotating the compass about the diameter of the leg of the landing gear significantly alter the determined front of the Phantom to eliminate toilet bowl and why?

    In my simplistic view, doing the compass calibration dance in a circular motion determines the center of the craft and the location of the physical compass, thereby determining the front of the craft. Magnetic north is the point of reference for the compass. GPS doesn't care.

    I warned you it was a noob comment...
     
  20. Fdnyfish

    Joined:
    Jun 6, 2013
    Messages:
    664
    Likes Received:
    36
    Location:
    Barnesville, PA
    Maybe You and Richard have a faulty compass