Welcome to PhantomPilots.com

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

Compass rotation

Discussion in 'Phantom 2 Discussion' started by tom3holer, May 21, 2014.

  1. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    I have the utmost respect for Ianwood and he asked if discussing compass rotation be in another thread so I am starting one.

    I live on Cape Cod, Ma and have about 15deg of Dec. We experience significant J hooking and TBE here with the Phantom.
    I have had 2 P2's and rotated the compass aprox twice the dec opposite the rotation as described in the Naza-2 Manual.
    In both aircraft it eliminated the associated errors. Today I tested my new P2 H3-3D with about 30deg rotation of the compass and it flew a full pack down to 22% with no change in flight characteristics during the flight. It flew straight with no j hooking and no TBE.

    True this is not a fix but it seems to work.

    If others have had good or bad experience doing this please post here so we can have a discussion of the pros and cons of this work around.

    Tom
     
  2. sar104

    Joined:
    Mar 22, 2014
    Messages:
    1,265
    Likes Received:
    245
    Location:
    Los Alamos, NM.
    I think a thread on this is appropriate, if only because there seems to be substantial variation in how well this alteration reportedly fixes those two specific problems.

    Just to clarify - having rotated the compass I assume that you did a compass calibration. If so, did you do the second stage with the Phantom pointing downwards as usual, or with the rotated compass pointing downwards? Having thought more about this I think that the former would be the correct method, since the IMU will expect to be in that orientation, but the compass won't know any difference - it will just detect a different vertical field component.

    I have rather less declination here (8.5) but I have noticed the hook effect without TBE, so I plan to try some compass rotation tests on one of my P2s for comparison when we get some calm conditions.
     
  3. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    Yes, I did do a standard compass calibration at the field before flying.
    I was not so surprised it worked and DID NOT correct the other way after flying a bit because it did the same thing on my previous P2.
    There has been a bit of talk about this being the wrong thing to do. It may be but it is the same procedure recommended by DJI on their Naza-M v2 to correct for the same problem.

    I would recommend being careful to make the compass secure and use fresh locktite on the screws.

    Hopefully, weather permitting, I will do more flying and testing tomorrow.

    Tom
     
  4. sar104

    Joined:
    Mar 22, 2014
    Messages:
    1,265
    Likes Received:
    245
    Location:
    Los Alamos, NM.
    Actually I agree with Ian that it can only be a partial solution, primarily because DJI indicated that declination correction is in the FC software. Had they not done that, and instead expected the compass module to be rotated, then it would presumably work fine, but the 2 x declination error strongly suggests that they included it but got the sign wrong in at least one place in the code.

    My initial reaction was that rotation should cure it, but the complicating factor of the FC learning process seems to make it less than ideal - even if you are not seeing that. However, it is not obvious to me that the benefits of rotation do not outweigh any current disadvantages - at least until we get a proper fix from DJI.
     
  5. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    Yes, I t is a partial solution but it seems to work for those that are adversely effected, like myself in a 15deg Dec area.

    I have had success doing this on three P2's. There was one instance that upon returning back it did not seem to track straight but it was before Mini Iosd and I do not recall there were any reds with the green flashes when I landed it indicating some Sat loss.

    I will be doing more testing, hopefully today, mainly to see if flight time will effect the calibration which I do not believe I have seen.

    I can't believe that believe that if indeed the correction is built in it is simply a sign change to fix it. Keep in mind the in the DJI line of controllers the Naza line, and the upscale WooKong both advocate rotating the compass to correct what we are seeing. Only the very top of the line A2 does not mention this.

    Tom
     
  6. sar104

    Joined:
    Mar 22, 2014
    Messages:
    1,265
    Likes Received:
    245
    Location:
    Los Alamos, NM.
    No question that other versions of Naza use rotation, but DJI obviously used a different approach with the Phantom - obvious if only because the compass installation is not amenable to rotation. Adding a database of declinations, or location-based declination algorithm, is quite trivial, and I'm sure that's what they did - hence removing any need to refer to rotating the compass. However, the symptoms clearly suggest a sign error in the code, especially the double-declination course error in course lock.

    It is, of course, possible that there is significant additional complexity to the problem, but since the sign error hypothesis seems adequate to explain the behavior, Occam's Razor suggests that we stick with it until further evidence demonstrates that it is not the correct explanation.

    The other puzzle is why some users report no issues in high-declination areas (are they just not noticing it), and why some users find the compass rotation to be an almost complete cure while others find it leads to the opposite problem developing during the course of a flight. That is still unresolved.
     
  7. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    Sar104,

    I agree with what you say with one caveat. If the coding is trivial why have they not included it in their other controlers? It would make life much simpler not having to do trial and error to find the correct rotation and changing it when you change location significantly. Changing it is a bit of a problem because you have to unstick the double sided servo tape to do it and after a few times the tape is useless.

    Tom
     
  8. sar104

    Joined:
    Mar 22, 2014
    Messages:
    1,265
    Likes Received:
    245
    Location:
    Los Alamos, NM.
    Good question. Hard to say with DJI - they are a most strange company, regularly contradicting themselves on the answers to seemingly simple questions, and making quite bizarre engineering decisions (such as the recent H3-3D design and modifications). They do not seem to have a very unified or well-thought out path forwards much of the time.
     
  9. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    Flew today with 3.04.

    Everything is the same as far as I can tell.

    I left my compass rotated about 25-30 deg to help with TBE and J-Hooking.

    I ran straight legs at fast to moderate speed and ran the battery down to 50%. Saw no change in tracking or J-Hooking. Would have run longer but my controller started beeping telling my batteries were getting low.

    Plan to fly later this afternoon and do more testing.

    Tom
     
  10. tom3holer

    Joined:
    Mar 20, 2014
    Messages:
    324
    Likes Received:
    2
    Location:
    Cape Cod, Ma.
    This is a copy of a post I just put up on the other thread on why compass rotation is a bad idea.
    DJI seems slow, to say the least, to get serious about this issue and there are those of us that need a solution, if partial, to fly our Phantoms without this problem.

    I understand the issues with getting DJI to correct the issue but this thread is about rotating the compass.

    I have done two more days of testing in my area of 15deg dec and what I found contradicts some of the reasons Not to rotate the compass.

    I rotated it about 25-30 deg CCW, the opposite direction of the TBE. At takeoff it has a very slight J-Hook. Within a minute or so it gone and by the end of the flight some 15 min later it still flew correctly with no change from the 1 minute period. I see NO overcorrection at all. The reports of it overcorrecting and going the other way I do not see in three different P2's, at least in my area. I seems to work exactly the same as my previous Discovery Pro with the Naza-M V2 in that regard.

    The caution is make sure you fix the compass in its new position so it will not change orientation during flight.

    I would like others who have tried this to post their results.

    Tom