Just built a large/heavy X8 config running 3.3.3 - 12 pounds and 17.5" props. Tiger 80A ESC and KDE 360 kV motors.
On the initial flight test the X8 would yaw left without input and even at full yaw input would barely rotate to the right. The log files show a pretty large difference in mean output between the CCW motors (outputs 1,3,5,7) and the CW motors (outputs 2,4,6,8) (see attached). I can't find conclusive information on this, but logically it seems like this should not be occurring. I'd expect all 8 motors to have output levels about the same when at a steady hover, is this accurate?
If so, what could be causing this? Obviously I've checked and rechecked the motor connections, spin direction and prop direction.
To clarify the difference in motor output, I think you need to look for an aerodynamic imbalance.
This is very normal for an X8 configuration with the same Propellers, Motors, ESC's - AND - with all bottom motors spinning one direction and all top motors in the other direction.
The bottom propellers experience a strong induced flow from the top propellers.
All ESC's, motors and propellers being equal, this causes a reduced amount of lift and torque for a given RPM.
Difference in lift is no issue since the overall lift of the four arms is still centered in the frame.
However the difference in torque has to be overcome, in order for the frame not to rotate along the vertical axis.
Therefore the bottom motors will have to spin faster, to generate the same torque of the top motors.
This causes the difference.
Normally a decent flight controller algorithm is capable of handling this difference, if this is not the case, try to increase the yaw PID gains.
To overcome this difference aerodynamically it is better to use propellers with a higher pitch on the bottom motors.
I hope this makes sense ;)
Please feel free to ask, if not.
@T, unfortunately your first assumption is incorrect. The prop direction looks like this: http://ardupilot.org/copter/_images/MOTORS_X8.jpg
I believe it was designed that way specifically to avoid the induced flow issue you detailed.
The X8 is set up the way it is to ensure that all axis have equal authority and are equally affected by the induced flow. That is why the X8 is basicly a H quad on top of an X quad.
The Y6 is setup to minimise coupling between the control axis. The original Y6 would not auto tune because this effect is so pronounced.
The main difference between the Y6 and the X8 is the Y6 will have throttle output separated between the upper and lower motors to balance the yaw torque. The X8 has the upper and lower motors always equal. This difference means the relationship between the upper and lower props and the induced flow is very different.
I just want to highlight:
The upper and lower motors of the X8 will be approximately equal when hovering. When applying yaw the pwm output to the upper and lower motors will diverge.
I am sorry I didn't make this clear in my previous statement. (thanks Stephen)
Thanks Leonard. I'm building a 3DR X8 frame and that's good to know.
"The X8 has the upper and lower motors always equal."
Well except for yaw behavior, yes?
Yes, If there is an acceleration in Yaw the upper and lower motors in the X8 will diverge to achieve the unbalanced torque. However, when in a hover the upper and lower motors should be equal. This simplifies the problem of optimising the performance because you can assume pwm output to the upper and lower motor will be equal.
In contrast, when in a stationary hover the upper and lower motors of the Y6 are not equal and this difference is taken up by the I term build up in the rate Yaw controller. If you change the combination of propellers the relative output to each propeller is dependent on the relative drag of each propeller. For this reason the best combination of upper and lower propeller for the Y6 will be different to that of the X8. It is also much more complicated to find the best combination as you have to simultaneously optimise both the power/velocity/flow and torque of both upper and lower propeller.
In addition to the motor direction shown here, the Pixhawk treats the X8 configuration the same as an X4 configuration, and sends identical signals to the motor pairs on each arm if I'm not mistaken.
I do not think so, it must spin top and bottom props differently for yaw.
Yes, duh I should have realized. In forward backward and side to side flight without yaw, the motors do act in pairs. Just looked at a few logs to confirm. In yaw they split up.