[chbot] Q-copter control -- was Unreasonable

[Mark Atherton]

Synco,

Yes, gryos are a good idea if you want to fly the unit manually. I 
got distracted before finishing my Q-copter, but some notes [you may 
find a better way, if so please report findings] :

Don't think the relationship between control signal and motor speed 
is too critical as long as "more" translates to "faster", and is monotonic.

Assume the motor control inputs below are called "pitch+", "pitch-", 
"roll+" and "roll-.

Two axis gyro needed (Pitch and Roll), with a controller in the 
craft. It is quite important to mechanically align the Pitch gryo in 
the correct Pitch axis :) make sure that the gyros are low gain (so 
can handle + and - 60 degrees without limiting).

Yaw can be done as a final mixing step, but you will need a third Yaw gryo.

Control loops and mixing and done on the craft, not the ground. Raw 
Speed, Pitch, Roll and Yaw joystick data is sent up from the ground.

For Pitch, the basic control loop [on the craft] is to keep the Pitch 
joystick data balanced with Pitch gyro output. Error is fed forward 
to slow down "pitch+" motor and speed up "pitch-" motor, and 
vice-versa. Gryo drift is automatically taken care of by the eye of 
the user flying the craft, by biasing the joystick. Re-calculate 
every 20ms or so.

Same code to process Roll.

Get Pitch working completely before trying to integrate Roll for 
multi-axis stabilization.

Speed control, something like

("pitch+" + "pitch-") / 2 = speed channel, with channel difference 
set by Pitch_Joystick - Pitch_Gyro.

("roll+" + "roll-") / 2 = speed channel , with channel difference 
being set by Pitch_Joystick - Pitch_Gyro.

Keep gyros away for the motors (magnetic interference) and draughts 
(temp shifts) - maybe mount inside a tiny polystyrene enclosure.

Your world will be different because of the Turnigy, but I modified a 
40MHz RC RX to output composite PWM and did the channel decode in 
software with the help of input capture hardware to measure pulse 
with. Gap is used to re-synchronize. Lots of error checking (repeat 
PWM cells within range of previous, and in the range 1-2ms) before 
accepting joystick data and of course of what to do with loss of 
joystick data...

4 ch joystick assignments: [left] speed and yaw, [right] pitch and roll.

http://www.idesignz.org/UAV/ has some incomplete notes

Regards,

Mark


At 11:59 p.m. 29/09/2011, you wrote:
Hey Chris,   Last week I also developed this need for a quadracoper 
so I slapped 4 motors, 4 motor-controllers, big LiPo battery and 2 
forward / 2 reverse propellers (else you have torque issues) 
together. Got all the mixing done in my new ER-9X transmitter which 
works nicely.... BUT I've concluded it's impossible to fly manually. 
You are welcome to come round and try :-) Believe you need to have 
motor controllers which are 'speed' controlled, (as opposed to PWM / 
torque control). Also wondering if you need gyros or whether you can 
fully stabilise it with a 3-axis accelerometer only? I've already 
reprogrammed a off-the-shelf brushless dc motor-controller (need to 
iron out some glitches), got a auto-pilot board with gyros and 
accelerometers etc... so only need some time to write software. Let 
me know if you are keen to help. /synco