Home Forums Everything about everything else YAW drift on Naze32/Flip32+ anyone having issues?

Tagged: 

Viewing 15 posts - 1 through 15 (of 20 total)
  • Author
    Posts
  • #12380
    Terje
    Moderator

    I’ve been battling yaw drift on my tricopters for a while now. yaw_deadband=5 and actually calibrating my Rx and FC has helped on the continuous drift. Turns out I have to tune every single (I have 4 now) Rx to the FC in the servo settings of the FrSky Taranis (to get 1500 as midpoint). Thought it was a tricopter issue, but my recent FPV trainer had the same issues.

    I now can hover in one place (indoors – no wind or other external influences) for 20-30 seconds. Then the craft will suddenly yaw drift clockwise. Sometimes correct back counterclockwise, sometimes just drift at an increased rate. I suspect magnetic interference, but have calibrated compass and magnetic inclination.

    Any suggestions? (apart from the noob comment: You should go with the original Naze32 bla bla bla.)

    #12383
    hsamuels
    Participant

    You are several steps ahead of me with the Naze32. But I wonder if the yaw loop even uses the magnetometer. It only needs the gyro, and it apparently works on Naze32 acro boards that don’t have a magnetometer.

    Perhaps baseflight recognizes and uses the magnetometer, which seems like a way for the performance to actually degrade.

    #12387
    donnyb
    Participant

    Hi all, I’ve had a similar problem with the yaw on my multiwii hex. I’ve read article where you can raise the p gains on the yaw and help stop it. It helped a bit but I also had to trim the radio. I flew my tri copter yesterday. Seemed like I could not get enough yaw. But still battled the drift. I’m using the flysky radio and receive. (Cheap) But I’m going to set everything up for frsky taranis. I’m going to work with the pids on the yaw.

    If you find out anything please post it.

    #12633
    Terje
    Moderator

    Another test on the Flip32+ today. First on my mini tri that hasn’t flown properly after I replaced the frame. Swapped the SBEC in the nose with a linear BEC to eliminate noise. Had no effect.

    Then on my tuned mini FPV quad. When exceeding 45 degrees of pitch, the craft will yaw after aprox. 25 meters. Same as on my tricopters. I have observed something similar on my gimbals that also use the MPU-6050 chip. Does anyone else fly beyond 45 degrees pitch? What are your experiences?

    #12634
    hsamuels
    Participant

    Good observation. If the scale factor of the three gyro axes are off, it could cause that problem. Does it seem to get even worse if you try to continue flight at the same pitch?

    If you are feeling ambitious, and have a record player, you may be able to measure the three gyros via Bluetooth. Orient the controller in each of the three directions.

    33 1/3 rpm should read 200 deg/sec.

    #12635
    Terje
    Moderator

    Say what now?!?

    🙂

    There is no way to continue the angle as the craft is slingshot into a spiral. I’ll post a short video tomorrow.

    #12638
    hsamuels
    Participant

    Yeah, that’s a problem. The question is whether it feels like the yaw is increasing faster than at a linear rate.

    The gyro has three axes. If you put the controller flat on a turntable (it doesn’t really have to be in the exact middle, but closer is better) then you will be able to read the sensitivity of the z-axis gyro. Of course you need a battery and a BEC, and also the bluetooth adapter on it, so it will look pretty funny.

    Then stand the board on one edge to read either X or Y, then read the other.

    I don’t have it all together here, but can you read the gyro output directly in one of the smartphone apps with the Bluetooth adapter?

    #12641

    Sometimes have luck with:

    *increase LPF (low-pass-filter), decrease Hz rate for servo
    *increase expo
    *slow or speed loop time
    *reduce or remove (I) for yaw
    *check receiver tab, adjust sub trim
    *reduce (P), increase (D)
    *increase ACC/Gyro filter?
    *(as you said) increase yaw deadband or neutral zone

    *small increments as too much remove control but not issue
    *fly more conservatively until something improves

    #12643
    hsamuels
    Participant

    After thinking about it a bit more, difference in sensitivity between axes can’t be the cause. I assume that you are careful about letting the controller calibrate the gyro offsets each time you connect power.

    Your description in the first post about the tricopter doing fine until sudden yaw drift after 20-30 seconds also seems to rule out offset calibration.

    So I’m out of ideas for the moment, other than just blaming the firmware (which isn’t out of the question).

    #12644
    Robert C
    Participant

    For the past few weeks I have been using a MPU6050 for a personal project and I needed to get reliable readings from it.

    I may be mistaken on some parts, but this is what I think is what is happening:

    Drift is one of the inherent problems of using a gyro. I have not looked too much into why this is, but you can only use a finite amount of digits, and when you are calculating with a finite amount of digits, you will always have an ‘error’, as it’s called.

    Well, why would you use a gyro, and not the accelerometers? Because accelerometers are very jittery. This jitter can somewhat be filtered out with a low-pass filter, but a better solution is to use a filter.

    There are several types of filters, the most used are a Kalman filter and a complementary filter. The basic thing is that the filters combine the data from both the gyrometer and the accelerometer to give the most accurate data you can get from the sensors. Depending on what you want, you can set the filters to depend more on the gyro or more on the accelerometer.

    For the roll and the pitch this works perfect, because the drift of the gyros can be corrected by the readings of the accelerometer. For the yaw however, there is no baseline. Only a magnetometer can be used, but that can be influenced by the wires running in your multirotor.

    So for yaw, there will always be drift. Maybe you have more drift than necessary, I don’t know, but there will always be drift.

    A good example is the following video:

    Credit goes to:

    Gyroscopes and Accelerometers on a Chip

    I hope this is a useful addition to the conversation.

    #12646
    Terje
    Moderator

    Thanks for the input Robert 🙂

    #12649
    Terje
    Moderator

    Here is a video of the flights. Running Cleanflight 1.7.2, but the problem is also present on 1.6.0 and 1.7.0.

    #12651
    Robert C
    Participant

    Ok, that is not normal yaw drift, that is just flying in circles 🙂 Very useful for Aerial Photography if you like NASCAR…

    Has this problem always been present? The board can take a beating but you would not be the first one with a damaged sensor. And due to the way the sensors are made, it can very well just affect one axis.

    I have flown with a pitch above 45 degrees and never experienced something like this, I can’t imagine that this is only a software fault.

    #12667
    Drehmo
    Participant

    It’s strange that your quad AND your tri do this. Admittedly, I’m still a noob, but I also have a 250 quad with a naze32 (w/ baseflight) and this doesn’t happen to mine. (*)
    If it was only the quad, my amateur opinion would be that the front right motor doesn’t get enough power at some point, or something like that?

    (*) though, currently not going beyond 45° because of angle mode…

    #12677

    Have you removed props and check live motor output with same throttle, pitch, angle? If not, maybe check throttle, then with pitch, then hold at angle, checking CF Motor/Servo Output tab. It might be the board or radio.

Viewing 15 posts - 1 through 15 (of 20 total)
  • The forum ‘Everything about everything else’ is closed to new topics and replies.