Fullscreen HTML5Med Res (??MB)HTML4

REMOTE CONTROL By BOB YOUNG Design factors for model aircraft This month, we will move on to a design study of a hypothetical R/C aerobatic aircraft. We will examine in detail the various design problems and the aerodynamic, mechanical and computerised options available to overcome these problems. To begin, we must have a clear understanding of the tasks we require the proposed aircraft to perform. Fig.1 shows the current FAI aerobatic schedule which is, as you can see, an awesome task for any aircraft. Compounding the difficulty confronting the model designer is the fact that, in competition flying, every manoeuvre is performed under the scrutiny of the judges. Those wishing to win cannot afford any shortcomings in aircraft design or errors in flying. This is a terribly demanding situation and just as full size motor racing has shaped the cars we drive today, the international aerobatic contest has shaped the models and radio equipment currently in use on club fields. The sad part about all of this is that the average flyer has little need for this level of sophistication but feels left out or deprived if his equipment falls short of this level. Expensive equipment does not make a good flyer. That comes with dedication and practice. Only after you have achieved complete mastery over the basic equipment and it begins to hold you back is it necessary to look for something better. Few ever achieve this level. I know that I certainly never did. It was always my flying ability that let me down. However I digress. Upon careful scrutiny of the F3A flight program, several interesting essentials emerge in relation to the design of our proposed aircraft. (1). The model must fly as straight as an arrow and perform all manoeuvres smoothly. (2). The model must perform all manoeuvres equally well, whether upright or inverted. (3). The model must be able to turn sharply. (4). It must be capable of flying on its side for some distance without loss of altitude. (5). It must be capable of vertical flight for some distance. (6). It must be capable of performing a snap roll. Going back to our basic aerodynamics, we can see that items 3 and 5 are complementary. Both call for a good power to weight ratio. Item 3 calls for a light wing loading as well. Item 4 is a complex issue but speed and power does help, again complementing items 3 & 5. Item 6 is a problem and is included for this very reason. This ·manoeuvre calls for large aileron deflections and is in direct opposition to the smooth flying required for all other manoeuvres. Here we arrive right at the heart of the need for the sophisticated encoders of today. Neutral stability There are two basic approaches to Item 6. One is to design an unstable aircraft that will snap roll with small control deflections. This leaves the pilot with the almost impossible task of flying smoothly in all other manoeuvres. The second is to design a stable aircraft and use large control deflections to overcome this stability. This is the preferred approach. In practise, we use a "neutrally stable" aircraft which is a very good compromise between stability and control. Neutral stability also gives us that "straight as an arrow" flight characteristic. The problem is, however, that while the centring accuracy of the servos is very good, it is not perfect and the large throws amplify this neutralising inaccuracy. Thus, the controls do not centre properly and the aircraft tends to wander, calling for constant corrections in flight. Looseness in the control linkages can add to this problem. Also, the controls become too touchy for normal flying and this shows up as a jerkiness in the flight pattern. Judges frown upon this sort of thing. A second complicating factor is that the aircraft controls are normally more sensitive around neutral and less sensitive as the control deflection is increased. All of this is, of course, precisely what the designers of this schedule had in mind. The old schedule had become too easy. They certainly fixed that. The old pattern was a much nicer, smoother routine. The way we set the controls for that pattern was to set full aileron throw to give the required 3 rolls in the 5 second time limit and full elevator to give the loop diameter required. This JUNE 1990 87 15 t 1 Oow,-nw11\Ct fflCl'I,"' I r LOOP WITH 2 50UAR[ rouR 1/2 ltOllS .,,, SIDEO TSl~Loor J .,_.. - • I I A Y'_ 19 - ~ 17 HUMPI\ BUMP W1Ttl OPTIONS l.,._,....I l 5 ) IMMELMANN MN 9 21 lUOl I I I J 1..- ~-, I J .,,so, ., .,,,,. ., -- .,,. ., .,,,,.;"✓' PROGRAMME St.ALL TURN WITH \ 2X 1/2 IOllu, ,,_ND DOWN ::::::::::::::: Fig.I: the current FAI (Federation Aeronautique Internationale) aerobatic schedule is an awesome task for any model aircraft. Illustration reproduced courtesy Model Aeronatical Association of Australia. resulted in a very smooth aeroplane in flight, with good centring accuracy on the controls. By contrast, the new pattern with its demands for square corners and snap rolls eliminates that approach and · has forced an electronic solution on the aircraft designer. Thus, by introducing a variable rate of control deflection at the transmitter end, it is possible to reduce the control throw to a more acceptable level for the smooth flying sections. The earliest approach tried was the dual rate switch. When activated, this switch gave a reduced rate of throw for full stick deflection; the percentage of travel being adjustable with an associated pot. But..! In effect, we gained little and introduced more complication. The SILCON CHIP ~ /-- F3A FLIGHT 13 HALF SQUARE LOOP WITH FULL ROLL IN VERTICAL I I 88 1/11011 vp, o, 1/4 rolh in bolh lt>gs ,;..,,,-"" centring accuracy remained unchanged because all that dual rate does is to reduce the pulse swing from 1.5 milliseconds ± 0.5ms to 1.5ms ± say 0.25ms. The minimum impulse of the servo remains unchanged. Minimum impulse The minimum impulse is the smallest pulse increment or change the servo can detect. Thus, if this figure is 5 microseconds, we will have 100 steps from centre to full deflection. However, if full stick throw at the Tx only delivers 50% of the servo travel, then we will only have 50 steps from neutral to full deflection. If we now increase the mechanical throw to double the control deflection to make up for the lost electronic movement, we have ef- -- -. .. · · londing zone IC>Oot . Ot 100,,, circl~ .·· ·· ~ ~ fectively doubled the centring inaccuracy. This is an important point to keep in mind when we are discussing ATV (Adjustable Travel Volume). However, the reduced throw on half rate does help make the aircraft a little less touchy. The trade off comes from the fact that there are now two separate sets of aircraft response times to learn - one for high rate and one for low rate. Flying is a terribly instinctive affair, for when travelling at 160km/h everything depends upon reflex action. There is no time to stop and think. The end result can be people starting outside loops at too low an altitude on low rate and being unable to reach the rate switch in time. The loop diameter is thus too large for the altitude available. This problem applies to all con- to as differential control and is a must on some aircraft. Finally, although the Tx control and servo potentiometers may theoretically be linear ("A" taper), they exhibit some non-linearity which will result in servos moving more in one direction than the other. The end result of all of this can be a most unsatisfactory combination of conflicting factors. Remember, the aircraft must perform as well upright as inverted. This means equal diameter loops, inside and outside; rolls equal in speed, left and right; stall turns and spins equally as precise, left and right. This cannot be achieved if the controls travel up more than down or left more than right. STICK DEFLECTION Mechanical compensation SERVO DEFLECTION Fig.2: servo travel vs. stick deflection for an exponential controller. In this type of controller, the servo travel progressively increases as the stick approaches the limits. trol functions fitted with dual rate and doubles the learning time. Exponential control A more sophisticated approach is to apply an exponential response characteristic to the Tx control stick. Fig.2 plots the servo travel against Tx stick deflection for an exponential controller. Here, the encoder electronics inodify the pulse output according to stick deflection. As can be seen, servo deflection is less per degree of Tx stick deflection around neutral but progressively increases as the Tx stick approaches the limits. This nicely compensates for the natural sensitivity of the aircraft around neutral but leaves the minimum impulse and centring accuracy problems unchanged. It does, however, eliminate the dual response time problem and thus to my mind is eminently more suitable for aircraft than dual rate or linear throw transmitters. Non-linearity But the complications do not end there. Most modern servos have on- ly a rotary output wheel to which the control pushrod is connected. This in itself is a nonlinear device again delivering less throw per degree of deflection as it moves from centre (Fig.3}. The same applies to the control horn - again, this is essentially a rotary device delivering a non-linear output (Fig.4). In fact, the output function of a rotating circle is a sine wave. This means that, at the extremes, there is very little change in throw for large changes in the servo angle. Admittedly, the non-linearity is small below 40° deflection but it is there nonetheless. The above explanation should make it obvious that it is incorrect to trim an aircraft for level flight with the servo arm off the 90° reference point. This will result in unequal control throws and loops or rolls of different diameters and speeds. One point here is that this nonlinearity can be very useful when setting up some controls, particulary ailerons which in some cases require only upwards deflection. This non linearity is referred Once again there are ways which do not rely on electronic gimmickry and we used all of these tricks for many years to good effect. Referring once more to Fig.3, it can be seen thatif the servo is travelling more in say the counter-clockwise direction (CCW) than CW, then by setting the servo neutral some degrees off centre in the CCW direction, we will get a good mechanical compensation for the nonlinearity of the servo electronics. As I said in an earlier column, smooth, accurate flying begins with the setting up of the controls and there are many ways to do this, not all of them electronic. This also has a secondary benefit which we shall soon see. But once again, there is a compounding factor. The aircraft may prefer to turn left rather than right or dive rather than climb. This is usually a result of poor design and may need to be taken into account when setting the control deflections. Now we are beginning to see why designers of radio control equipment are constantly searching for more flexibility in their encoder designs. The microprocessor is ideal in this situation. Computer encoders The modern computer encoder has many features which allow us to compensate for the large number JUN E 1990 89 75' 90" 60' Fig.3: the servo output wheel is itself a nonlinear device which progressively delivers less throw per degree of deflection as it moves away from the centre position. CONTROL SURFACE \ 0 0 0 TO SERVO Fig.4: because it is a rotary device, the servo horn is also a nonlinear device. This means that at the extremes, there is very little change in throw for large changes in servo angle. of non-linearities encountered in rigging an airframe. For example, adjustable travel volume (A TV), in which the travel each side of neutral can be adjusted, is ideal for compensating for non-linearities in the electronics and airframe. Exponential Tx controls are very useful for compensating for the natural sensitivity of aircraft around neutral, and the nonlinearity of the rotary output wheel and servo horn. In fact, using rotary servos without exponential control is quite wrong, even though we did it for years and will continue to do so. I am expressing a purely theoretical viewpoint here. In practice, we somehow manage. In fact, when you consider that we could fly a nice pattern with a reed set in which all we had was neutral and full throw, all of this really is nit-picking. I was just so glad to get a basic no-frills proportional set that even now all of this gingerbread is meaningless to me. I still feel no need for it and continue to fly with a basic no-frills 5-channel set. Endpoint adjustment Endpoint adjustment (EPA) or 90 SILICON CHIP ATV is used for overcoming nonlinearity in the Tx and servo electronics. A separate potentiometer is provided in the Tx to allow each endpoint, CW (clockwise) and CCW (counter clockwise), to be adjusted. Computer encoders use a key entry to set the percentage of throw. Thus, the control deflections can be set precisely equal about the centreline. If required, they can of course be set up unequal to compensate for aircraft control characteristics. This is an extremely important feature and quite safe to use, unlike servo reversing which is potentially hazardous. Once you have set up one aircraft to a Tx, it is really dangerous not to have all subsequent models set up the same way. If you have to reverse one or more controls before flying that second model, you are really tempting fate, especially if one of those controls is the ailerons. One nice thing with the computer encoder over the old balanced voltage types is the fact that some transmitters are fitted with a memory which can be programmed to retain the servo trims and travel directions for each aircraft. Up to six aircraft can be stored in some Tx. This at least avoids accidents involving reversed controls, provided the correct program is selected for the aircraft being flown. Yet the story does not end here. Just buying an expensive set does not solve all of the problems. We have still not dealt with the problem of minimum impulse. This must be dealt with in a more subtle way. Good servos needed First, for aerobatic competition, you must buy high quality servos. These feature ball bearings on the output shaft to minimise bearing slop, minimum backlash gear trains for centring accuracy, coreless motors for low current consumption and short transit times. They also feature very precise electronics which have a good minimum impulse figure. Now we are ready for the real work. To begin, we must now turn to the control geometry. Modern model aircraft are fast and place high loads on the control surfaces, as do modern model cars. Exactly the same considerations apply with regard to control geometry. Control flutter Control flutter can be encountered in almost any model aircraft. This is a situation in which the control surfaces vibrate at an extremely high frequency in resonance with the airflow. The noise is just like a "bullroarer" and I have learned to cringe when I hear it. This is an extremely dangerous problem and can vary in its effects from ripping the teeth off servo gears to tearing the control surface completely off the model. I have landed many a model with controls in shreds after being hit by control flutter. It most commonly affects ailerons but all controls are vulnerable. The cure is very stiff linkages, hinges and horns. Any backlash or slackness in the system will allow this problem to manifest itself. Now we arrive at the real implications of using less than the full servo travel available. Fig.5 illustrates the basic geom- SERVO ARM l J___:-=- MOST RIGIO, LEAST MOVEMENT 'V' OSCILLATING FORCE OF AIR . . .l ARM LEAST RIGID, MOST MOVEMENT Fig.5: maximum rigidity is obtained when one end of the pushrod is close to the servo bearing and the other is on the outside end of the control horn. Carried to extremes, however, this gives an unacceptable reduction in control deflection. HINGE c----------i----------) (a) (b) ELEVATOR SECTION, SENSITIVE NEUTRAL SYMMETRICAL SECTION , SOFT NEUTRAL etry involved. Maximum rigidity is obtained when the pushrod on the servo arm is close to the servo bearing and farthest away on the control horn. The problem here is the reduction of control deflection as we carry this to extremes. Thus some compromise is called for, the essential point being that all available servo travel should be used to achieve maximum control rigidity and accuracy. EPA (end point adjustment) works by reducing the available pulse-width deviation Fig.6(a) shows an elevator section that will be sensitive around neutral but less sensitive at the extremes. By contrast, Fig.6(b) will have a soft neutral but increased effectiveness with increasing deflection. available from the transmitter encoder and thus also reduces the servo travel. As useful as it is, over zealous use of EPA on flying controls will only rob you of system performance and open the way to other nasties. However, EPA really comes into its own on throttle, where the end point adjustment is extremely important. One last point on the problem of control flutter. The full size practice is usually to mass or aerodynamically balance all control sur- RCS Radio Pty Ltd is the only company which manufactures and sells every PCB E, front panel published in SILICON CHIP, ETI and EA. 651 Forest Road, Bexley, NSW 2207 Phone (02) 587 3491 for instant prices faces. In aerodynamic balancing, some surface area is placed in front of the control surface hinge line to provide some damping. Mass balancing calls for the weight of the elevators to be balanced by an equal mass placed in front of the hinge line. Sometimes a combination of both is used . Finally, having gone through all of the above to ensure that we have a nice equally responsive aircraft, we find after take-off that it is a dog: sensitive around neutral and very reluctant to roll in one direction. We land and scratch our head. The other problems must wait until the next few columns but the control sensitivity is bound up in the shape of the flying surface and the control surface itself. Before computer encoders, we used this fact of aerodynamic life to introduce exponential control or vice versa. Fig.6(a) shows the cross section of a tailplane which will be very sensitive around neutral and less sensitive at extremes. Fig.6(b) shows one soft around neutral but with increased effectiveness as deflection increases. There are a lot of factors influencing this situation and it is very difficult to design the control response predictably, hence the usefulness of exponential control and the computer encoder. However, when the correct design is arrived at, the aircraft becomes very pleasant to fly. I had one aeroplane that was so accurate on elevator control that altitude could be controlled to within 2cm. This allowed me to land that aeroplane on full throttle on a moderately smooth surface. Some aircraft cannot hold ± 10cm and are very vague to fly. In conclusion, the computer encoder is a very useful tool, but must be used with a studied approach to the problems involved. Used in a casual or lazy manner, it will give no better results than the old encoders. With that we must end for this month. I think I am in big trouble with the Editor. I said I could probably cover aircraft design in three issues. So far we have discussed control deflection; only about 150 more factors to go! ~ JUNE 1990 91