Hugh Hemington
Well-Known Member
As I wait for parts, I'm thinking of the job the controller has to do, and thinking about how this job is made harder or easier by the geometry of the frame, which spaces the arms and motors.
Logically, it seems like a perfect 90deg. X (or plus) where the pivot point and motor spacing are identical is easiest to control, since power applied to one axis rotates the craft exactly on the axis of the opposing "cross" arm. And since the arc of rotation on all four arms is identical, as the quad is rotated, on that axis, no mixing is required to continue to produce the same correction. And inversely, the farther from the same vector the arms are, power applied to rotate in one axis also rotates to a degree in the other.
Following this logic it seems that the more the vectors described by each opposing arm deviate from parallel and/or the SAME vector, the harder the quad is to make stable with motor adjustments because the axis of rotation each produces is more oblique to frame-center.
Further, it seems logical that the chip on the controller board which measures tilt on any axis needs to be located as close as possible to the center of the vectors described by the opposing motors. (ideally also the opposing arms) That way, adjustments directed by the controller have a LINEAR effect on axial tilt.
I realize most controllers have settings for 'H' vs. 'X' vs. Plus etc. to compensate for this, but I'm assuming there is an ideal form for simplicity, and every other form will invariably introduce some error despite controller programming. In short, the less compensation the controller needs to do, the less impact it can have if it does it poorly.
It seems like the most problematic layout would be straight front and forked rear arms.
Right now, my frame is about an inch out of square. the motor centers are one inch different front to side, and the pivot points are also one inch wider than long. This will probably have little impact on flight, but there is still time to shorten the frame across the front, making both the pivot points and motor centers describe perfect squares, and bringing opposing arms onto the same vectors.
What are your thoughts?
Logically, it seems like a perfect 90deg. X (or plus) where the pivot point and motor spacing are identical is easiest to control, since power applied to one axis rotates the craft exactly on the axis of the opposing "cross" arm. And since the arc of rotation on all four arms is identical, as the quad is rotated, on that axis, no mixing is required to continue to produce the same correction. And inversely, the farther from the same vector the arms are, power applied to rotate in one axis also rotates to a degree in the other.
Following this logic it seems that the more the vectors described by each opposing arm deviate from parallel and/or the SAME vector, the harder the quad is to make stable with motor adjustments because the axis of rotation each produces is more oblique to frame-center.
Further, it seems logical that the chip on the controller board which measures tilt on any axis needs to be located as close as possible to the center of the vectors described by the opposing motors. (ideally also the opposing arms) That way, adjustments directed by the controller have a LINEAR effect on axial tilt.
I realize most controllers have settings for 'H' vs. 'X' vs. Plus etc. to compensate for this, but I'm assuming there is an ideal form for simplicity, and every other form will invariably introduce some error despite controller programming. In short, the less compensation the controller needs to do, the less impact it can have if it does it poorly.
It seems like the most problematic layout would be straight front and forked rear arms.
Right now, my frame is about an inch out of square. the motor centers are one inch different front to side, and the pivot points are also one inch wider than long. This will probably have little impact on flight, but there is still time to shorten the frame across the front, making both the pivot points and motor centers describe perfect squares, and bringing opposing arms onto the same vectors.
What are your thoughts?