I have previously called attention to the fact that engineers sometimes come up with bad ideas. I can't claim to be immune from this problem. Confession to a couple of projects that didn't turn out exactly as planned may therefore be appropriate. Fortunately, no great harm was done, and I learned some lessons from these experiences.
Improving on the Wright Brothers
When the Wright brothers first made their gliding experiments at Kitty Hawk, their biplane gliders had no vertical tails. They soon found that attempts to make turns by warping the wings were unsuccessful because the drag of the wing on the side twisted to increase the lift would cause the glider to swing around in the opposite direction from the desired turn. This phenomenon has since been called adverse aileron yaw. The Wright brothers then tried mounting rudders behind the wings that were linked to the wing warping controls so that they would deflect to offset the adverse yaw. The result was immediate success in producing good control in turns.
Later, the Wright brothers and most other airplane designers gave the pilot independent control of the rudders so that the human pilot could use the rudder as desired to offset the aileron yaw and could also use the rudder for other purposes such as ground control, landing in crosswinds, and spin recovery. Nevertheless, elimination of a separate control for the rudder remained a subject of interest because it was thought that correct use of the rudders was a difficult problem for novice pilots and that airplanes to be flown by the general public should be as simple to control as automobiles. Airplanes in which the rudder control was eliminated were called two-control airplanes. Perhaps the most notable examples of airplanes of this type were the early models of the Ercoupe designed by Fred Weick.
A problem with the direct linkage between the rudder and ailerons as used by the Wright brothers is that the amount of rudder to be used varies with the airspeed. This problem was not serious for the Wright brothers because the speed range between the stall speed and the maximum speed was very small. This variation arises because the adverse yaw of the ailerons increases directly with the lift coefficient. As a result, much more rudder deflection is required to make properly coordinated turns at low speed than at high speed. Human pilots can approximate the required variation by applying the same force to the rudder pedals for a given aileron deflection at any airspeed. On manually controlled airplanes, the rudder deflection produced then varies inversely as the square of the speed, exactly the variation required to....
.....give a yawing moment that increases directly with the lift coefficient.
One method to cause the amount of rudder deflection to decrease as the airspeed increased would be to operate the rudder from the aileron control system through a spring. The hinge moment applied to the rudder would then be independent of airspeed so long as the rudder was in neutral. When the rudder moved, however, the spring force would fall off due to the tension on the spring being released. As a result, the rudder deflection produced in low-speed flight would be less than that required, whereas in high-speed flight it would be more nearly equal to that required.
To avoid this problem, the restoring moment applied to the rudder by the interconnecting springs should be offset by an unstable spring moment that would tend to deflect the rudder away from neutral. I reasoned that a simple way to provide this effect would be to sweep the rudder horns back. I had occasionally used this technique in the past to vary a restoring moment from stable to unstable. I built a small model to demonstrate the principle. A sketch of this model is shown in figure 17.1. In this model, a small control stick was connected to threads that ran through pieces of curved tubing to act as guides. Then springs were attached to the ends of the threads and attached to the control horns on the rudder. By sweeping the horns back the correct amount, the rudder would stay in any position, which indicated that the restoring moment of the springs had been offset by the unstable moment produced from the spring tension being applied aft of the hinge line.
The model worked perfectly, and it was therefore decided to try the system in a small low-wing trainer that was available at the hangar, a Fairchild PT-19. A photograph of this airplane is shown in figure 17.2. The installation was very simple and required only that new rudder horns be built and that springs be attached between the control cables and the rudder horns. In addition, control of the rudder was transferred from the rudder pedals to the aileron system. A sketch of the control system is given in figure 17.3.
When the system was installed, a disconcerting effect was found. With the airplane on the ground, the control stick would not stay in neutral, but would immediately slam rather violently to one side of the cockpit or the other. The unstable moments applied to the rudder destabilized the entire control system! This effect had not been noted on my model because the friction on the threads in the guide tubes had been enough to keep the control stick from moving.
After some consideration, additional springs were attached to the control stick to offset the unstable spring moments, as shown in figure 17.3. Then the airplane was considered safe to fly. Flight tests showed that the rudder coordination in turns was satisfactory, but the aileron control forces were excessive because the aileron control stick had to supply the force to deflect the rudder as well as the ailerons. Because the rudder.....
....had been designed to operate with rudder pedals, to which the pilot can apply considerably more force with his legs than he can apply to the control stick with his arms, the rudder hinge moments increased the lateral control forces by a large amount. This problem could have been alleviated by redesigning the rudder to be more closely balanced aerodynamically, but the work involved in making this modification was not considered worth the effort.
A brief theoretical analysis showed that despite the provision of centering springs on the control stick, the control forces on the ground are unstable about neutral over a certain range of deflection. The reason for this effect is that the rudder, lacking any restoring tendency, goes to full deflection as soon as the control stick is moved a small amount. The springs in the rudder system then apply a force to the control stick causing it to move away from neutral until the springs on the control stick balance this decentering force. As the airspeed increases during the takeoff run, this instability disappears.
A draft of a report on this project was prepared by Stanley Faber, an engineer in my branch, but the report was never published. Figures 17.2 and 17.3 are taken from this report.
As a result of this experiment, I learned that my crude models could not always be trusted to give the right guidance for full-scale design. The program also recalled the reasons that the Wright brothers had abandoned the two-control arrangement; namely, the other uses for the rudder (primarily, landing in crosswinds) that make it inadvisable to eliminate this control. Fred Weick retained the two-control arrangement on the Ercoupe airplanes that he produced, but when another manufacturer resumed production of the.....
....Ercoupe, the control system was changed back to a normal three-control system.
Relearning the Laws of Angular Momentum
A frequent objective of flight research is to determine the stability derivatives of an airplane; that is, such quantities as directional stability and damping in yaw. This procedure could be greatly simplified if some means were available to apply external forces or moments to the airplane in flight. In the early days of the NACA, the damping in roll of a Curtiss Jenny was measured by dropping sandbags of known weight from the wing tips and measuring the response of the airplane. This sort of procedure is not practical on modern, high-speed airplanes, however. Most research on measuring these characteristics has been conducted by measuring the response of the airplane following a displacement of the controls and using a mathematical process to fit the response with a set of derivatives that produce motion closely approximating the measured response.
Nevertheless, an idea occurred to me to build a device that would be capable of applying oscillating moments of varying frequencies to an airplane in flight. A sketch of this concept is given in figure 17.4. Basically, the device consisted of a small flywheel that was spun back and forth at varying frequencies by an electric motor. Then the torque produced by the flywheel was magnified by running it through a gear train with a large reduction ratio. For example, a 10 to I reduction would multiply the torque by a factor of 10. If four stages of this gearing were used, with larger and stronger gears and shafting on each stage, the torque could be multiplied by a factor of 10,000. It appeared that the torque could be made large enough to actually oscillate the whole airplane in flight.
A device of this type was built and tried in a fighter airplane. On returning from the flight, the pilot said that he hadn't been able to detect any disturbance to the airplane.
Obviously, something was wrong. It didn't take much thought to realize that the torque applied to the airplane had to be equal to the change of angular momentum of the small flywheel and gearing, a rather negligible amount. The larger torques involved in the gear train were offset by reaction torques in the framework that held the gears, so they had no external effect. My face was rather red when I explained this situation to the pilots and the engineers involved.