EN CN
IMU faulty?
  • mkp
    Is my IMU (or something else) faulty? It seems that when I roll the IMU, the yaw axis responds. The further I roll the worse it gets. Past about 45 degrees and it starts oscillating. Pitching it does influence it a little as well but doesn't go into crazy oscillation. I know that a little cross axis interaction is normal, but this seems excessive. Thanks for any help.
  • AOPEN3434AOPEN3434
    no you need to set the orientation of the sensor in the GUI so when you tilt up the arrow moves up you roll right the the roll moves right . Set X, Y, Z sensor orientation
  • mkp
    Thanks for the reply. Sorry, maybe I wasn't specific enough. The gimbal is generally working fine and has even been used on production so far. I'm trying to improve the performance of it since I'm having problems at more extreme angles. I don't mean that it responds completely incorrectly (definitely would notice that), but it seems there is some interdependent motion going on that I can't sort out. It's very slight up until the point where it causes oscillation. At this point I'm not certain that it is in fact the IMU but perhaps something further down the line. Either way, the same scenario still exists.

    My fear is: I know the IMU can't be placed at the nodal point for all three axis (since there is a camera there), but perhaps this causes some inaccuracies that lead to problems like this.
  • mkp
    Theory (that applies to my other underlying problem): The closer one gets to the extremes of travel the closer we get to a condition where two different axis line up. As a result we get to a sort of feedback situation where two motors achieve the same effect.

    For example: at 90 roll, the yaw and tilt motors effectively perform the same action.

    In between these extremes there is a transitionary state. You can see the yaw influenced for example when you move the tilt and roll motors (easier to see when those two motors are off). The system doesn't necessarily know which axis to correct and could go into a feedback/oscillation state.

    Is it possible that this is the cause of why the gimbal oscillates the further one pushes it from the centred balanced position?
  • mkp
    Furthermore: it seems that the YAW measurement in the IMU works in any orientation and appears to "right itself" in the unit. Am I correct in that assumption? If that wasn't the case then it appears this feedback situation would be minimized since it would measure only the Yaw relative to the actual Yaw motor.
  • kikojiu
    It is true but I don´t think anything can be done. There are angles in which the axis "cross". It think is around 35 degrees.
  • mkp
    Thanks for the confirmation. I guess technically where the axis "switch" would be is 45 degrees, but there would be immediate cross axis interaction as soon as things are out of balance. So that means tuning not only has to balance the typical adjustment performance within the axis, but also has to be tuned to a sort of "asymmetrically over dampened" state to prevent this within reasonable limits. The more level you can hold the gimbal in operation, the less you have to do that. My tuning experiments seem to indicate that in real world use as well. Would be nice to hear from others (especially basecam) to confirm this concept.
  • kikojiu
    I would prever they found a way of limiting it. I hate it when we pass that limit and the gimbal is shaking all over. Professionally doesn´t look good. It looks the electronics are faulty.
  • mkp
    For anyone following this thread or coming across it due to similar experiences. It turns my suspicions were correct and this is expected behaviour. Also found out that this seems to be addressed by the new 3-axis board with the ability to use a second IMU. The gimbal community is amazing but also often a little guarded so it took a while to find confirmation of this (though I kinda enjoyed thinking through the problem). There's a post on the 500+ page rcgroups thread that makes reference to it. Hopefully this will save someone else some time and research.

    Essentially, be careful to not tilt the frame too far. You can't tune that particular oscillation out (though like I said, you can try to maximize the reaction of this behaviour a little bit).
  • b_digital
    Thanks
    is this related :
    I have an issue right now, when i am tuning the pid i take the rool and give a ~30deg tap so i can monitor it. but the more i roll the unit the more the yaw take a bit of angle !!!! clockwise and conterclockwise. It is like it wanna to compensate with the Yaw for the roll.
  • mkp
    Quite possibly yes, but there's another situation where a similar thing happens as well (that I've never experience but just read about so can't speak from experience). Just out of curiosity how do you have your IMU oriented?
  • b_digital
    upside down so i can read the pinout for the wire
    i am also having the wobble when i go to far tilt up or down
  • b_digital
    another thing on my side is, i did try to calibrate the acc the best i could
    but i get a good level horizon (ex: when looking north but if i look east it tilt one side and if i look west it tilt the other side :(
  • mkp
    The conclusion that was confirmed by basecam was that you cannot tilt the frame too far in lock mode. This is a limitation of the current boards (with only a single IMU accessible).

    In follow mode this is compensated for.

    Are you in follow mode?

    However, it sounds like you have some other issues. Calibration appears to be one of them. All I can say for that is to repeat the process as accurately as possible. Sorry, I can't tell if anything else is wrong. Maybe someone else can help on the forum.

  • kikojiu
    shaking on tilt extremes is absolutely normal. But also seems you could use six position imu calibration.
Add a Comment