I'm sure this has been discussed here before, but past posts seem pretty well buried now.
I've been giving some thought as to whether rudder or aileron (in isolation of each other) is better for initiating turns via autopilot control. As a bank-and-crank RC flier (it wasn't until relatively recently that I learned that the left stick on my transmitter could be used for more than two things, idle and full throttle), my initial thought was to use aileron. But the more I thought about it, the more I realize that rudder is probably a better way to actually turn a plane with anything less than a 60º bank. (Just in case anyone reading this is not familiar with sport RC flying, the elevator actually makes the turn when the plane is in a steep bank. If rudder is used at all, it's used in the opposite direction of the turn just to keep the nose up. Indeed, some jets started coming out without any rudders at all, just fixed vertical fins. The AMA, I believe, now has a rule that functioning rudders must be present.)
So, I'd be interested in hearing what others have to say. Right now, I believe ArduPilot will give you either rudder or aileron, not both (correct?). If it gives both, it seems it would be based on a single, but mixed, signal. Would there ever be a need or benefit to having independent rudder and aileron control? I'm sure full-scale UAVs have this, but it might be a bit too much for an amateur UAV.
Thanks,
Paul
Comments
BTW the monster I keep in my hands is 2.4m wingspan rudder only, 5KG max AUW, bungee takeoff - and then there are no cables in the wings at all, only 2 screws to bolt and go! Thanks to well chosen weight distribution it flies pleasantly.
So this is the upper limit, because Cularis with rudder only is not fun at all to fly (roll response comparable to.. dreamliner. believe me it only sounds good).
Also, I don't want to give misleading information. The AMA rule I was referring to applies to turbojet powered model airplanes. Unless I'm mistaken, they must have functioning rudders. As it happens, I'm building a ducted fan delta-wing model without rudders.
I am flying a Senior Telemaster, so this is a conventional looking airframe with a tail. High wing; flat bottom airfoil, tractor engine. This airframe exhibits adverse yaw in turns. I've added a bit of differential aileron throw, but it's still a pig through corners and doesn't track very well.
I have been computing a target bank angle to steer myself to the next waypoint. I use aileron to chase the target bank. I've posted videos recently here in other discussion topics so I won't spam them here too. The glass cockpit videos I've posted that show all the subtle interactions between control surfaces, attitude, speed, and autopilot target values are already a bit dated anyway.
In full scale aviation, pilot's are taught to "step on the ball" ... the little ball that slides back and forth gives an indication of lateral acceleration + gravity. This is actually important for tracking through turns correctly. Otherwise you might skid out or slip in ... maybe analogous to a hook or a slice in golf. Mostly you will probably oscillate and "dutch roll" through the turn rather than doing anything consistant which looks bad, especially from the on-board perspective. It can make the airplane look like it is fighting itself.
In FlightGear we model the "slip/skid ball" using the side to side acceleration (ay) divided by the vertical acceleration (az). Using vertical acceleration in the formula is to model how a real ball would behave. But what we really want to do for controlling the aircraft is to zero the lateral accelerations. So I've been playing around with a PID stage for my rudder that tries to drive lateral acceleration (ay) to zero.
I'm using an ardupilot for servo control, so this means that my throttle (ch3) and rudder (ch4) are driven from 8 bit timers and pretty crude/coarse in their output. My concern is that the step size of the rudder might induce more lateral acceleration than it fixes. I've seen this when I turn the gains up. That said, I did some flights yesterday with the turn coordinator turned on and the gains dialed down really low and I think I like the results. With a better servo controller I should be able to do even better. I don't have hard data, but perceptually, the aircraft tracks straighter through the turns.
Another thing I've noticed with my turn coordinator turned on is that the airplane self trims. It's hard to know if your rudder is exactly trimmed with your ailerons or if you are achieving straight and level flight through some slightly cross controlled (but balanced) configuration. However with independent aileron and rudder control, the ailerons can drive the wings level, the rudder zeros out the lateral forces, and you end up (I believe) with a nearly optimal straight and true trim configuration. That's kind of neat.
I would probably disagree with a couple nuances in Krzystof's reply, but I'd rather share my own thoughts.
I wouldn't go out of my way to add rudder to a flying wing or "jet" style design that has fixed vertical stabs. I wouldn't go out of my way to add ailerons to a rudder only design (easy glider, etc.) But given a choice of airframes, I would want both.
So in the end I think the optimal result is *highly* airframe dependent, and probably highly mission dependent too. And like anything there are many ways to attack a problem and good engineers make their chosen approach work well.
Rudder only:
-the only surface working during flat landing approach or takeoff run
-minimal setup, eight, power, price, faster assembly
-the less surfaces you have to tune, the smoother they interact, shorter tuning/retuning, more reliability
-allows removing minor course errors without banking, but theoretically losing some speed (the lost is not noticeable nor easily measurable).
-gives automatic antispin protection
Ailerons only:
-very effective after launch in manual mode
-dangerous for automatic flying close to stall speed (changes airfoil shape)
-gives more agility, never fully used in smooth UAV flying
-easier to tune to give 'average' results
-much easier to tune for fast flying at high G
-more servos, more cables, longer assembly
Aileron+Rudder:
is apparently mixed with ailerons for almost all known autopilots, it is more worth optimizing the engine setup than minor rudder deflections in presence of antennas and parachute boxes etc.
-too much constants to tune
-nonlinear interactions
-more servos, heavier
-LESS reliability (any servo failure IS final unless you fly it and demonstrate otherwise with state-of-the art research experimental self-tuning autopilot, good luck with that)
In short I think you are trying to optimise 'because real pilots do that'.
Note in real flying the rudder is kicked in during entering the turn, then slowly retracted a little.
It is very difficult to tune across range of speeds and angles, and without real market advantage.
It is not even fun to do.