/dist/images/branding/favicon

Roll Radian goes off level when Multi pans

Discussion in 'Radian' started by John Gore, Oct 19, 2012.

  1. John Gore

    John Gore Member

    Joined:
    Jul 24, 2012
    Messages:
    93
    Likes Received:
    15
    Just noticed an issue. When I pan my MK OctaXL with 2-axis gimbal around, the roll goes off level. The more I pan, the more off level it goes. Once a full rotation is done, and the OctaXL is held facing that direction, the radian is perfectly level again. Odd.

    I have my roll Radian powered by a 7A BEC, and no inputs (it runs independently).

    Could it be that my Radian is mounted "skew" ?

    Perhaps it needs to be calibrated or something? (have only played with gains, but so far found that the standard 100% seems to work best all round for my 7D, cx760 is a little different, but still to tweak that)
     
  2. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,383
    Likes Received:
    1,164
    Hi John:
    That's very odd.
    Does this happen when you have the MK OctaXL suspended above the table, or only when its in flight? I suspect JohnC is going to need some additional diagnostic information.

    Can you take an image of installation of the Radian sensor, please, and upload it?

    Thanks
    Andy.
     
  3. John Cunningham

    Joined:
    Jul 5, 2012
    Messages:
    183
    Likes Received:
    13
    The first thing to check for this is that the Radian is mounted level to the roll axis and perpendicular to the pan axis. The effect you describe is normal for an IMU system if it does not have accurate angles. Check this first before we dig in any deeper ...
     
  4. Pavlos Antoniou

    Joined:
    Oct 6, 2012
    Messages:
    321
    Likes Received:
    18
    i think that level to roll and perpedicular to pan axis is the same think right???! Or i am missing something..! Iam ready to install mine thats why i am asking :) :)

    I just updated them to the last version! :)
     
  5. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,383
    Likes Received:
    1,164
    Level to roll means horizontal. Perpendicular to pan means vertical.
    [Edit: that "level to roll" bothers me....perhaps it's me that's missing something...]

    So, not quite the same thing. :) [Edit: At least I don't think it is.]

    Andy.
     
  6. John Gore

    John Gore Member

    Joined:
    Jul 24, 2012
    Messages:
    93
    Likes Received:
    15
    Thanks guys, have been on the road a lot, just got back home now, will try get down to resolving this soon.

    Some more info:
    -- The roll is not off by much. Maybe 2 degrees at worst point, and only noticeable when flying higher than the trees and buildings so a complete horizon is visible.
    -- I do not yet know when it is worst, but I think at 90 degrees from starting direction. So if the Octa is facing "left" or "right" from the starting point).
    -- I did not notice this problem in the beginning, only noticed it it later flights. Not sure if its a new problem, or was always there. (likely it was always there, but I was so impressed by the Radian stabilisation I did not notice it moves slightly off level).
    -- I am pretty sure I have mounted the Radian correctly, as I took great care to mount it level, and square. On roll it is off by less than 0.5 of a degree (leaning right), but I have not even bothered to correct that in the Radian software yet as it is so minor.

    Quick question. Does the Octa need to be on a perfectly level surface when the Radians are powered up? (It is usually not on a perfectly level surface).
     
  7. John Cunningham

    Joined:
    Jul 5, 2012
    Messages:
    183
    Likes Received:
    13
    John,

    The Radians don't need to be level when they are powered up. They automatically work out what angle they are at.

    2 degrees is a small amount for an IMU like this - but looks like a lot in the video. The symptoms you describe can happen if one axis in the Radian IMU is slightly off. The ideal mounting is where the Radian tilt axis is exactly in line with the camera lens for the tilt axis (even though it is correcting roll) which in turn means the roll axis is exactly in line with the camera roll.

    If they aren't you can get apparent errors where tilt transposes into roll (and vice versa) as the IMU rotates. They completely swap every 90 degrees - which seems to be what you are seeing.
     
  8. John Gore

    John Gore Member

    Joined:
    Jul 24, 2012
    Messages:
    93
    Likes Received:
    15
    Tks JohnC. Just to see if I understand correctly:

    If the camera is tilted down 40 degrees (tilt), then the roll Radian and camera lens are no longer similar on the tilt axis, so the horizon going off at 90 degrees is more likely to happen. But when the camera is level, and roll Radian is in line with it on tilt axis, then the horizon going off at 90 degrees is no longer (or much less)?

    I am wondering how much the little bit of flex in the double sided tape used to mount the Radian effects this.
    Also, would vibrations or wobble effect the ability of the Radian gyros to "read" position? (I am working a a small amount of vibration which I should have sorted soon)
     
  9. Pavlos Antoniou

    Joined:
    Oct 6, 2012
    Messages:
    321
    Likes Received:
    18
    If its level to roll it will be and vertical to pan and the opposite...

    Thats why i am saying its the same..:):)
     
  10. Andy Johnson-Laird

    Andy Johnson-Laird Administrator
    Staff Member

    Joined:
    Jul 31, 2012
    Messages:
    10,383
    Likes Received:
    1,164
    Is this how you're interpreting the axes PaNt? (The roll axis in this image that I stole from quadrocopter.com is not completely level, but it illustrates the point I'm wondering about):

    Level to roll, perp to pan.png

    I think we may need JohnC to clarify what he means if it's not the above.

    Andy.
     
  11. Pavlos Antoniou

    Joined:
    Oct 6, 2012
    Messages:
    321
    Likes Received:
    18
    Yes if one is true then the other is true tooo..! Thats it..!

    Also on the tilt radian the best you can do is to loosen the boom of the titl and turn it a little so the carbon plate of the radian is leveled to the tilt..! so it will be a lot easier to install the radian :) :)
     
  12. John Cunningham

    Joined:
    Jul 5, 2012
    Messages:
    183
    Likes Received:
    13
    Andy/John,

    I'm just saying that the more accurately the 3 axes of each individual radian module are aligned with the camera lens, the better the stabilisation result should be. For example, if they are not quite aligned then a tilt module might also sense some roll input when the camera tilts.

    The thing to remember that even though a Radian is only stabilising one axis, it is actually processing the gyros from all 3 axis in order to calculate the pitch+roll angles in the internal IMU.

    I wouldn't expect a bit of vibration to affect the overall accuracy of the angles from the accelerometers - but it might bring in some errors from the gyros depending on the amplitude and frequency. The main issue here would be if it results in noise to the output to the servo motor : if it has noise then it might fight against the inertia of the camera.
     
  13. John Gore

    John Gore Member

    Joined:
    Jul 24, 2012
    Messages:
    93
    Likes Received:
    15
    Hi guys. Thanks for all the input, and sorry for not replying sooner. Only managed to test a few days ago again, after making a few changes to the multi.

    Here are the results I am getting now (raw footage, only cut boring bits out).

     
  14. Tabb Firchau

    Tabb Firchau Administrator
    Staff Member

    Joined:
    Jun 28, 2012
    Messages:
    911
    Likes Received:
    472
    Looks great John!
     
  15. John Gore

    John Gore Member

    Joined:
    Jul 24, 2012
    Messages:
    93
    Likes Received:
    15
    Thanks Tabb. Did you notice at 2:30 and 2:47 the horizon is not level for some time, but later corrects? Hmm...
     
  16. Tabb Firchau

    Tabb Firchau Administrator
    Staff Member

    Joined:
    Jun 28, 2012
    Messages:
    911
    Likes Received:
    472
    Hi John, it looks level to me at 2:30 and 2:47. It looks like you are looking off into a valley of some kind and their are hills on either side. If I average the 'horizon' in my mind I think it is level.

    Tabb
     
  17. Tabb Firchau

    Tabb Firchau Administrator
    Staff Member

    Joined:
    Jun 28, 2012
    Messages:
    911
    Likes Received:
    472

Share This Page