The problem I have is that while my Pixhawk is sufficiently dampened from vibs my camera gimbal is possibly resonating with from them. This problem is observed in flight by a violent shaking left and right while the gimbal shakes fore and aft.
In "paint mixer" flying mode, vibrations can be observed on the IMU GyrX and ACCy log from the pixhawk. The multi shakes left and right while the gimbal bounces back and forth (possibly up and down as well). Log file linked & picture attached.
I have tried a few different things:
-Softening the PID values on the pixhawk this results in a quad that is barely flyable. The violent shaking is gone but the video footage is still jumping up and down on the gimbals pitch axis. (Various gimbal tuning does not remove camera shake)
-Adjusting the PID on the gimbal has no effect on reducing or eliminating paint mixer mode. Even with the gimbal powered off but mounted the shaking is reduced but not gone.
-Without the gimbal attached the multi fly's well but still oscillates slowly on pitch and roll.
-Without the multi the gimbal works well
-The gimbal is raising the COG of the craft from below the pixhawk to directly through the pixhawk.
-The gimbal weighs 750grams, it is attached via a damping plate rated for 800 grams. other plates have been tested ranging from 800-2400gram damping
At best through various PID auto tune attempts (With and without gimbal attached) / combined with manual tuning (Significant increase in P term well above autotune value .1 increased to .22) It oscillates slowly ~1hz (about 1 time per second) on the pitch and roll axis while an audible almost too high D term noise comes from the motors. Lowering D has no effect on removing the noise, Lowering P results in further instability. Rate D is currently set to .004. The gimbal still shakes although not as much on the pitch axis seen as a vertical bouncing on recorded footage.
Also, an audible wind up of the motors can be heard periodically throughout the flight in calm winds. This noise has been described as I term windup but i do not understand the concept enough to begin its troubleshooting. During the windup the craft is relatively stable, but seemingly unresponsive / sluggish. If I increase the IMAX values would this result in an out of control flight condition?
In short i'm looking for somebody to help / who has experience with:
-Tuning when gimbals (alexmos related?) and multi's don't work well together
-I term windup and IMAX effect?
-Removing slow oscillations on pitch and roll
-Somebody to explain why my Rate Yaw P only works when significantly above max value. Currently at .7 with rate yaw I at ~.007. Lower values result in yaw oscillations. At its current value it holds heading well. It just doesn't seem right but it works?
At this time I can only offer a means to efficiently mix large paint containers weighing < 5000grams
Side note:
Using FFT's and an external vibration analysis tool I was able to determine my multi is generating vibrations in the ~200hz range. Multiple harmonics of this base frequency can be seen propagating along at a substantially reduced amplitude. this could be a possible cause of the gimbal making the multi not function well but I am not sure.
About the frame:
X configuration
6S 5A Battery
KDE 515kv motors
12" carbon fiber props
20A Afro ESC's
pixhawk running 3.3 Rc8
16mm CF tubes.
AUW 3500grams
Max thrust
~9000
About the gimbal:
Total weight ~750grams
Alexmos 3axis carbon fiber gimbal
damping plate
Camera platform balanced across pitch, roll and yaw axis. no movement of camera regardless of position.
Link to log file;
https://drive.google.com/file/d/0By12f3-1EZywMDhjSlcwM2E4Yk0/view?usp=sharing
Replies
How much effort have you spent balancing your motors and props? That's the first line of action.
I have added a short video showing the Z axis bouncing from an onboard camera's perspective. This shaking is beyond what the gimbal can compensate for. Using the alexmos software the I term is 0.01. Any higher and the shaking becomes worse.
Link to 5 second Video:
https://youtu.be/YWJJc5ZSIKA
Vibration Video
Software stabilized version of video can be seen here:
https://www.youtube.com/watch?v=W4EPqraCBQQ
Software Stabilized version
I will attempt to lower the parameters associated with Z axis acceleration to compensate tomorrow and see how it goes.
Scott, Thank you for taking the time to respond.
-All of my props have been balanced and doubled checked with 3 different prop balancers and 2 different people. Admittedly some of the CF props required more sanding than i've ever seen before but the problem persisted across 2 different sets of blades.
-The motors from KDE are amazing. When attempting to dynamically balance them nothing could be done to improve the balance on the motor. Any additional weight in any location increased the vibrations measured in my testing rig. When they are running you can place your hand on the motor arm and feel almost nothing.
Images regarding Pixhawk reported Vibration below:
-Since my initial post I have removed the damping plate and then 3d Printed a thick flat plate and hard mounted the gimbal directly to the frame. With this action the paint mixer effect was completely removed but the slow oscillations persisted. I have never had a gimbal / frame combination work better without a damping plate between them (Vib balls of various resistances.) more testing is required to understand the why this is occurring.
-Once the violent shaking ended I started to re-tune each axis with auto tune / manual tuning in an attempt to remove the oscillations. After the Auto tune completed the multi continued to oscillate on both the pitch and roll axis roughly 1-2 times per second, slowly.
The PID Values from auto tune on pitch and roll continued to be surprisingly low .1 KP and .01 KI with the D param slightly higher up to .055 from .04. With the new tune values applied I used chan 6 knob to manually adjust the Kp with a range set to 0.1 - 0.4. At the end of 2 flights the final Kp value ended up at 3.53.
- All in all I believe this to be an improvement but there is still room for further adjustments as the oscillations persist but at a slower rate. I am slightly concerned with the significantly Higher Kp values but they seem to work well without any feedback wobble. In order to fly such a responsive multi i needed to lower my ATC_ACCEL_R/P_MAX values significantly to ~20000 while keeping yaw at 15000. I am curious as of to what happens when these values get to low.
-At the end of today's testing / tuning / hair pulling experience this was the end result for vibrations from the log file
The vibrations appear to have improved drastically on the X and Y however the Z axis vibrations are of particular concern to me:
Also of note the vibrations shown from the VIB tab in the log look far worse but are they? what do these mean?
Link to log file from flight:
https://drive.google.com/file/d/0By12f3-1EZywc3BRVUpPY0VubWs/view?u...
Log file from flight