Octa quad - error in the algorithm

Now diagonal with motors 2,5, 4,7 work for throttle control and 1,6, 3,8 for YAW control. BUT, it’s work, as traditional quad - completely the same level throttle for coaxial motors - primitive mixer.

Traditional quad control yaw by reactive moment, but on my octa almost does not exist reactive moment on each arm - no reaction to yaw conrol.
And what we have: if bottom more powerful, than top - yaw axis reverse, if one top motor disabled - rotate by yaw in one side and crash.

Log: 2017-12-14 11-44-14.bin (618.9 KB)

Can you explain a bit better. We fly multiple octa quads and the yaw is responsive.

Yes, if you have same motor on top and bottom, and this botoom motors spin free

yes, I do have the same motors on top and on bottom.
I also use the same propeller size on top and on bottom.

But I temporarily used 10’’ props on bottom and 11’’ props on top, and still had yaw responsiveness.

What do you mean with “bottom motors spin free” ?

Same bottom motors do not create a lifting force. I have more kv on bottom and thay are work at same power with top motors. Your bottom motors work with little power and not create reactive moment for compensate top motors, so standart quad mixer can work with your octa.

I suspect you have not read the code and understood the way the mixer works. I think this is a classic case of blaming the code for a poorly setup aircraft.

This is completly incorrect.

I have done tests on coaxial propellers using different pitch, diameter and kv on upper and lower motors. All have responsive yaw control and all can handle multiple motor loss provided no two are on the same arm and if there are at least two cw and ccw motors left then yaw control is maintained.

Looking at your log I suspect your aircraft is poorly tuned and under powered. If you have asymmetric yaw response I suspect you also have some mechanical issues this is backed up by the 10% yaw offset you need to keep the heading from changing.

I look in log on pwm output and see, how it works in practice, not in code

I am sorry to inform you but your interpretation of what you saw was incorrect.