EN CN
Roll loses horizontal angle randomly
  • OctovisualsOctovisuals
    Hi, friends.

    Yet another question I have in regard to my 8-bit 2.40 b7 one-IMU BaseCam controller. I know this has been discussed several times before, but I think it's good to talk about it the more the better since it's a very critical issue.

    My gimbal sometimes loses horizontal angle randomly, even if I recalibrate gyro constantly. In the land, I always recalibrate gyro before every flight, the roll is perfect, but in flight sometimes it gets to one side and very slowly returns to 0 (it returns faster when it loses steps, for example, with I = 0.02). This causes a loss in time and flights since the shot can't be used. It's random because it can work perfectly all the time, but then it loses horizon a couple of times a day.

    I think it's not low Power since it holds well and doesn't lose steps... I don't think it's temperature, since I recalibrate gyro before every flight, so even if it's very hot, it should recalibrate considering that temperature, correct? Could the accelerometer have anything to do, even if it's recently calibrated? Could some HDMI cable 1 cm above the IMU affect it electromagnetically, or something else nearby?

    What could be the causes of this misbehavior? What could I do to fix or prevent this effectively?

    Any assistance will be much appreciated.
  • OctovisualsOctovisuals
    I think it's the IMU, when it's affected by temperature... When it's calibrated in a temperature that differs notably from the one when working with the gimbal.

    Can we assume that when the roll isn't level when the gimbal is static it's gyro's fault, and when the roll drifts in movement it's accelerometer's fault?
  • Garug
    I would more say vice versa, but I do not think it is that simple overall.

    Best to have both well calibrated.

    Temperature for sure affects. It is good to calibrate the gyro at the temperature the gimbal is used.

    I think accelerometers are not that critical for temperature.

    There is also the temperature calibration procedure. it works, but is somewhat difficult to perform, I usually do not do it.

    I find calibrating ACC well during installation and performing gyro calibration when needed (temperature changes, location changes or when there is any problems) sufficient.

    I have assigned gyro calibration to a button and normally skipping it during startup.

    Gyro calibration when needed is easy. just press the button and leave the gimbal steadily on floor.

    Note, the IMU takes time to settle to the ambient temperature. allow at least some 10 minutes before calibrating when the temperature changes.

  • OctovisualsOctovisuals
    Thanks for the reply, Garug.

    I always try to have them both precisely calibrated. I usually recalibrate the acc. everytime I use the software (once per month at least), and always recalibrate the gyro almost before every use (with the radio, when the gimbal is on the ground).

    And it's usually good level, but occasionally the horizon loses some degrees, usually when the temperature is hot (like in summer on the sea...). And it seems to happen mostly when used with a drone. That's why I asked if the acc. is affected by movement in relation to roll degrees...

    That's what bothers me, not knowing exactly what causes the behaviour, despite doing everything I know to avoid it. Fortunately, it only happens like three times a year... But it's not professional when working for a client, and makes me lose time and battery charge.

    Maybe there's more here than acc. and gyro calibration. But I also considered this already...

    Well, I'll keep testing and investigating anyway.

    Best regards.
  • Garug
    Small and slow roll variations will easily happened because lateral accelerations when flying. Luckily roll is easy to level on editing SW.

    Experimenting with different gyro trust settings could help reducing it.

    Here is a very clear example how lateral accelerations affect the roll, starting at 0:10
  • Mizamook
    I am reading with interest after fighting with temp. Acc. cal.

    So does it work, or does it not? If it does not work, why is it there?

    Roll is very important .. resolution loss due to correcting in software is bad for my mood.
  • Garug
    On well build and setup gimbal Roll will work good on normal conditions, but high and long lateral accelerations will cause temporal roll and/or pitch drift depending of the acceleration direction. This is with any gimbal that does not have advanced GPS tracking etc.

    Normally it is well avoidable when recognizing the problem and taking it into account when shooting video.

    On normal handheld work you should not see any of it.
  • ariellago
    all very nice, but the problem is solved when you have a 32-bit controller. So the problem is not GPS, but software or hardware.
    I had the same problem with controllers 8 bits, until I get tired of calibration.
  • Garug
    It depends what the "problem" is. Both 8 bit and 32 bit can provide steady horizon on normal use and on both of them horizon will drift on long lateral accelerations, like on the above video. The above video is with 32bit + encoders.

    32 bit could have some improved algorithms though to compensate lateral accelerations (I do not know), but it has the same basic problem.
  • anaka
    Hi people i have exactly the same problem of the video
    i have 32bit thiny board and i use ti on a drone
    also if i fly slowly when i move lateral respect to the gimbal i slowly loose the horizon on Roll

    i have encoders and all is well balanced and calibrated and this issue is present by year 2013 so from 8bit boards

    googling around this is a tipical limit of all gimbals

    on this video there is the solution with mavlink so i'm asking how to correctly get the same results?
    i have mikrokopter 2.5 flight control

  • Garug
    Good imu calibration helps, well performed temperature calibrated provides pretty good results. However even very good IMU calibration does not remove the problems caused by accelerations.

    Connecting to flight controller can provide better results. FC that has magnetometer and GPS has better information of the actual horizon level than the gimbal has.
  • Garug
    Is that a two axis gimbal? If not there is some serious probelmes with yaw. Pitch seems to have some problems too.
  • anaka
    Hi Garug i meant the vudeo in your post...a bit above mine

    i have 3 axis gimbal and generally speacking it works pretty well except for this little issue
    the horizon lost start to be noticeable in some conditions only, it happens with accellerations but it's not that exagerate as in the video in my post

    i'll try to increase the Gyro Thrust as suggested in some post above and give more feedbacks
    maybee it could be enough in my case

    one more question is: can i use the mavlink connection with mikrokopter? do you have any info on this?
  • Garug
    On the above video you posted, yaw is not performing well, maybe you have too tight follow yaw setings.

    I have tested the FC connection, but not Via Mavlink, I guess that should work too. I would suggest using the latest Beta FW or at least never FW. you are using very old version 2.59
  • Garug
    There is no magic number of gyro trust that will make it good. You need to test and experiment, also you can make it adjustable via RC and use different gyrotrust in different flight situations. When high gyro trust is used, it is important that gyro is well calibrated, rather temperature calibrated.
  • anaka
    thanks for your tips Garug

    once again the video in my post is from AlexMoskalenko and not from me, i posted it just to show that someone did the fix with mavlink ;)

    BTW i'll follow your tips and write back my impressions, i have the lates enocder FW nd have to say that is really better respect to the past ones

    what about the FC connection you are talking about? does it works like mavlink? how to use gps data in that case?
  • Garug
    Ok, I misunderstood as your video.

    There is still the FC connection via PWM, tested it long time ago, and yes it provided some improvement, It gets the horizon from FC (DJI NAZA servo gimbal control outputs were used), but too complicated, small horizon drifts are easy to correct during editing.
  • anaka
    Ok Garug

    i'm on the MK forum as well asking the same questions, if they have a solution will update this thread as well

    thanks again
Add a Comment