I recently strengthened my carrier board mount on a Tarot X4 by installing a piece or 3mm carbon sheet. The goal was to reduce vibes - and it’s succeeded at that. (They still need to be reduced a bit further.)
This modification has reduced the copter’s vibes over all - but they still need to be reduced further.
On a test flight today at 2.5, 5, 7.5 and 10 meters per second ground speed, clipping was recorded on IMU-2 - the one that’s hard mounted in the Orange Cube. IMU-0 and IMU-1 showed no clipping - even thought the their reported vibes were slightly higher.
This graph shows a relationship between ground speed and vibration. I’ll be working to address this issue this week.
Does EKF3 look at IMU-2 - and register it’s vibes and clipping? If I can’t eliminate the clipping, is there anything I should do so that EKF3 won’t swap lanes because of it?
I believe hard mounting the carrier board is the method advised by CubePilot and the ArduPilot community. Is there anything I should consider doing to help IMU-2 avoid clipping?
It’s a different IMU (part number), so its measurement range is very likely different than the other two. Thus it may clip at a different value. A look at the datasheets for each IMU will likely bear this out and make your observation make sense.
It is unfortunately not that simple.The vibration dampening of the Cube boards is tuned to a particular frequency band. If the vibration frequency of your system has high amplitudes outside of that particular dampening band you should add external dampening to the cube.
But … be careful, your external dampening method might actually amplify the vibrations if it is not tuned to address them. To do this you need to consider the mass of the cube and the stiffness of your damper. This is not easy to do. Hence the recommendation to just hard mount it.
PS if you find some usable damper that has a stiffness datasheet, let me know about it. I have not found any yet.
Fair enough Amilcar - I may have spoken a bit too soon and oversimplified.
Vibes overall are pretty rough on that graph. First step is prop balancing, and I think you’ve asked how to do so in another topic - I don’t have a good answer for folding props. I often use 3M mounting tape as a thin damping layer when mounting Cubes. The stuff I linked is quite strong, and I trust it for long term mounting without hardware. I find that it often helps but doesn’t get so squishy that IMU performance suffers (worth a try when hard mounting gives poor results). I also take a lot of care to tie wires as securely as possible, leaving very little slack (but not “tugging” on the connectors).
I never iso a Cube with the exception of 3M tape. I hard mount everything. Vibrations can be simple as a bad prop. The other day I had a small vibration and it was a off center prop doing the issue. They make a plastic standoff that screws on the fame. The advantage is air can pass under the Flight controller. especially nice if you have mRo Pixracer pro and want to hardmount it.
I use a lot of 3M VHB tape - I have it in three different widths in my tape box. Really good stuff. I mount quite a few parts using it. Right now I have my carrier board attached to the mount with the four 2.5mm screws that the ADSB carrier board supports. Switching to VHB may be an interesting experiment.
Before changing the mounting, I’m going to try the carbon fiber straight props that I’ve ordered. At least I can balance those. If I still have vibes with props that I know are balanced, I’ll go back and try other carrier board mounting options.
Thank you @amilcarlucas and @Quadzilla - it’s always appreciated when developers contribute to a thread.
While the mounting options and considerations are one aspect of this problem, the other is understanding how EKF3 handles a situation like this where IMU-2 is clipping and IMU-0 and IMU-1 are not.
As I recall, EKF3 was the product of a graduate student’s work under @tridge . Would it be possible to have someone familiar with EKF3 comment on how it will respond to a situation like this - and if there are any options in the parameters that may head off potential problems?
For example - if it’s known that IMU-2 will clipping, would be be possible (or advisable) to disable it? Or maybe change it’s clipping threshold?
Typically in Ardupilot there is one EKF lane per IMU, the clipping IMU should cause its lane to have poor confidence and therefore not be used for navigation. So it shouldn’t cause a crash unless other bad stuff happens to the system, but if you have a case of bad lane affinity you could have a problem.
IMHO vibrations are still to high. I would try stiffening the plate so it doesn’t act like a speaker diaphragm amplifying Z vibrations.
