Well, remember that there is software out there that will calculate a custom motor mix for you. There's even an online calculator that will let you do it graphically and cut/paste the mixes in any MultiWii-based firmware.
It was my experimentation with the KK2.xx flight controller and then moving to newer FCs with more advanced Accel modes that spurred my own research and brought about this understanding about how the two halves of the MPU actually work.
As far as my use of the "crutches"... well, I have "started over" 3 times in my process of "Learning to Fly"; the latest linear progression from Angle to Horizon to occasionally Rate modes has produced, at least for me, much more successful stick time and less frustration. I was already flying in stabilized modes with "toy quads" for a long time before I even knew what "rate" mode in a quad was.
Crashing daily and proud of it,
mnem
"...Tongue-tied and twisted, just an earth-bound misfit... I."
Yes, but it's the math behind those mixer calculators that's pretty flawed, and what I was pointing out above. The relationship ends up being totally non-linear due to the r^2 factor in the moment of inertia calculations (totally ignoring the interaction of the other variables that complicate it even further), and then the mixer is not being applied by the FC in a way that makes the impact as expected. As Boris pointed out, it's actually a hold over from the way some old PID controllers did things that aren't even used anymore. The only real benefit to the mixer right now is making your PIDs more similar, adjusting for vectored thrust (A-Tail or V-Tail) or remapping motor outputs. I'd have to check in more detail to see how they're being applied to the accelerometer, but I don't believe it's really any different. Again, it might be helping but it's mostly blind luck IMO, and the fact that we happen to be in a range where the changes aren't that far off from linear.
Also an important point here, as you hit on, it's the distance of the accelerometer from the CG that seems to matter to level modes, not the distance of the motors from the center of rotation, which is all the custom mixers are accounting for.
mnementh:
Aug 27, 2016 at 09:53 PM
Well, remember that there is software out there that will calculate a custom motor mix for you. There's even an online calculator that will let you do it graphically and cut/paste the mixes in any MultiWii-based firmware.
https://www.iforce2d.net/mixercalc/
It was my experimentation with the KK2.xx flight controller and then moving to newer FCs with more advanced Accel modes that spurred my own research and brought about this understanding about how the two halves of the MPU actually work.
As far as my use of the "crutches"... well, I have "started over" 3 times in my process of "Learning to Fly"; the latest linear progression from Angle to Horizon to occasionally Rate modes has produced, at least for me, much more successful stick time and less frustration. I was already flying in stabilized modes with "toy quads" for a long time before I even knew what "rate" mode in a quad was.
Crashing daily and proud of it,
mnem
"...Tongue-tied and twisted, just an earth-bound misfit... I."
quadmcfly:
Aug 27, 2016 at 10:10 PM
Yes, but it's the math behind those mixer calculators that's pretty flawed, and what I was pointing out above. The relationship ends up being totally non-linear due to the r^2 factor in the moment of inertia calculations (totally ignoring the interaction of the other variables that complicate it even further), and then the mixer is not being applied by the FC in a way that makes the impact as expected. As Boris pointed out, it's actually a hold over from the way some old PID controllers did things that aren't even used anymore. The only real benefit to the mixer right now is making your PIDs more similar, adjusting for vectored thrust (A-Tail or V-Tail) or remapping motor outputs. I'd have to check in more detail to see how they're being applied to the accelerometer, but I don't believe it's really any different. Again, it might be helping but it's mostly blind luck IMO, and the fact that we happen to be in a range where the changes aren't that far off from linear.
Also an important point here, as you hit on, it's the distance of the accelerometer from the CG that seems to matter to level modes, not the distance of the motors from the center of rotation, which is all the custom mixers are accounting for.