Thanks for responding Bill.
Yeah my first thought was also that the blades where not extending straight out, when simply pulling outwards while gripping the tips of the blades. So I actually spun up the rotor head on the helicopter, to get them perfectly straight, and then carefully removed the head and shaft, making sure to not hit the blades and then placed it on the balancer.
Anyway, I’ll call it good for now and try to fly it again hopefully soon.
Hi Oliver,
I just watched your video and saw something you might have overlooked. The outer part of your swashplate is rotating when you turn the rotor and then the heavy side of it is slowly moving downwards and accelerating your rotor. That doesn’t allow a precise measurement.
Nevertheless I don’t believe this is causing the effect you are seeing. It definitely looks like the rotor blades aren’t straitened out perfectly. They might be in a straight line when the rotor is running, but as soon as you shut down your engine, they will leave their position. I know that, because I’ve installed lead-lag sensors in a helicopter and wanted to calibrate the zero-position. Not possible like that. I had to average the position of the flight measurements.
Hi Felix,
Thanks for responding.
I did notice that the part of the swash that goes into the anti-rotation bracket would rotate downward slowly. I have tried letting the rotor blades go while holding onto this piece by hand and it gave the same results.
Interesting about the blade lead and lag. It makes sense that they would leave the straight line as soon as the rotor slows down. Didn’t think of that.
How can you then ensure that the blades are completely straight out from the blade grips. I have tried holding onto the ends of the both blades while pulling outward, and it balanced the same as when I spun up the rotor system on the helicopter.
A little update.
I got to fly it earlier today with the blades balanced and tracking well but the oscillations were still present on the roll axis.
So I turned on the notch filter and set the frequency to 42 which I determined from the logs.
The settings I used were.
INS_NOTCH_FREQ 42
INS_NOTCH_BW 10
INS_NOTCH_ATT 30
INS_GYRO_FILTER 10
INS_ACCEL_FILTER 10
And this seems to have gotten completely rid of the issue.
Below is a plot of the vibes during that flight
My question is whether or not this will cause problems with the EKF as the vibrations are still present, just filtered away. I believe I read somewhere that the EKF doesn’t use the filtered gyro data.
Any input is greatly appreciated.
/Oliver
@Oliverl96. Those vibes are really good! I think you should raise your gyro and accel filters to 20 and leave the notch the way it is. If your rotor speed is 42 hz then I think having the ins LPFs at 20 would be fine.
Also those vibes should not cause any issues for the EKF.
Nice work!
Hi @bnsgeyer,
Thank you so much for the continued help.
Alright, I will raise the cutoff frequencies to 20 again.
I do have one more question already. When looking at the desired and actual rates for roll, pitch and yaw, it is clear that the FF needs to be tuned as the amplitudes are not matching, but it also seems like there is an offset between the two. The wiki says this can be caused by poor CG or the swash not being level correclty. Should I look at leveling it again or try to tune the FF values first?
Again thanks alot.
/Oliver
That is true but the integrator will take care of that. You need to set your ILMI to 0.08 for pitch and roll and that should get rid of the bias.
ILMI for both pitch and roll is set to 0.08 but the I gains are at 0 at the moment. I will get to tuning then as soon as I have at chance to get out and fly again.
I don’t think there is an (easy) way of moving the blades in a perfectly straight position. The hard way would be to build some kind of frame that you could align to the long straight lines of the leading and trailing edges of the blades. But I don’t think you’d really need that. As Bill already said: Your vibrations look pretty good. After the balancing mine looked pretty much like yours:
But the thing I noticed: Your rates look way noisier than mine. This are the rates of the same flight I showed you before. You can also see that offset, because I had my I gains at 0 back then. Just like yours.
So I have a theory what might have caused your problems: Would it be possible, that your controller mount is too soft and the controller is constantly oscillating around on its base? I also started with a too soft mount and the helicopter almost tipped over on the ground before the rotor had spooled up completely. No way of getting it into the air like that. My rates of that ground test looked like that:
Okay, if the notch-filter completely solved your problem. Great! Just wanted to let you know my thoughts.
Hi @Felix,
The original plots were without the rotor system balanced so the oscillations were huge.
After balancing, the oscillations were basically gone on both pitch and yaw but were still present on the roll axis. This could well be due to too much soft mounting tape underneath the flight controller. I do also have the servo outputs directly soldered onto the pixracer, and the wires run along the roll axis of the helicopter over the flight controller, (even though they are strain relieved and have room to move), so it could also be because vibrations are transferred from the frame into the wires, and then finally into the flight controller that way.
I’m not entirely sure. For now I think I will call it good and use the notch filter.
Thank you for taking the time to type up a detailed post.
/Oliver
Hi Oliver,
I think it is normal, that you see higher vibration levels in the roll-direction, than in the other axis with a traditional helicopter. That is because of the much lower moment of inertia of the roll movement compared to the pitch movement. So I think you don’t have a problem with the wires coming in from the side.
I was a bit confused, that you are talking about yaw-oscillations. I think that might be Z-accelerations. But I’m not sure, what “vibe” actually means (accelerations or rotations). If you want to know it precisely, you can look into the IMU data and see Gyr X, Y, Z and Acc X, Y, Z.
Yes, if it’s working that way, leave it as it is. Never change a running system.
That’s what the ArduPilot Community is for 
