His calls to the pilots brought no response. The bombardier climbed from his compartment in the Fortress and made his way to the cockpit. To his horror he found pilot and co-pilot slumped in their seats, badly wounded and unconscious.
Pulling the pilot from his seat, the bombardier took his place and turned a few dials. The mighty bomber swung into a wide turn, straightened out and headed for England. Three hours later it settled for a crash landing, its crew safe but for the wounded men.
This miracle of flight was accomplished by a new automatic pilot whose story now is being told for the first time. The success of every bomber raid today is due in part to this device which is linked with the Norden bombsight.
This autopilot makes four important contributions to AAF bombing raids. It holds the plane on an unerring course. It relieves pilots of fatigue by taking over their duties for long periods of time. It has saved many lives because of its ability to fly the plane even after the controls have been shot away. Most important, it gives the bombardier a steady aiming platform.
In a trip like the bombing raid on Bangkok, Thailand, from bases in India, for example, Maj Harry Watkins led his squadron through to the target in a flight lasting 16 hrs 30 min. Sixteen hrs 10 min of the time he used the autopilot. The 20 minutes of manual control were used in taking off and landing.
Capable of more than 300 flight corrections per minute, the autopilot, manufactured by Minneapolis-Honeywell, holds a plane rigidly on course during the bombing run, when an error of just one mil would cause the bombs to miss the target by more than 360 feet.
Use of the Minneapolis-Honeywell autopilot has reduced precision bombing mean error by 50 per cent. Thus, if the mean error is reduced from 1,000 to 500 feet, the bombing effect is quadrupled.
This means that nine planes can do the work of 36. Return trips to the target are cut to a minimum, fewer lives and planes are risked, hence fewer are lost. Materially, fewer planes and men must be provided; logistically, fewer planes and men need to be maintained. Result is an incalculable saving in men, money, planes, materials, transportation and time.
The M-H autopilot is similar in principle to the automatic pilot used by commercial airlines, though more precise. The gyroscope remains its basic instrument.
The autopilot differs from the familiar Sperry gyropilot, which is used on commercial airlines, in several ways. It is fully electrically operated. Its servo motors connect directly with the surfaces being controlled, whereas the Sperry pilot controls the surfaces by small hydraulic pumps operated by servo motors. The Minneapolis-Honeywell pilot has a follow-up system connected with the amplifier, while the Sperry follow-up is connected with the flight gyro.
Long before the human mind is aware that anything is amiss, the autopilot has countered the effect and corrected the deviation from the set course. This is done by two gyro controls. Contacts on the gyroscopes pick up weak electric signals which are amplified and sent by wires to the "servo motors" operating the ailerons, rudder and elevator. The process is reversed when rough air moves the control surfaces.
One gyro control handles the horizontal or directional motion, the other the pitch and roll movement. Sensitivity of the gyroscopes is so acute that the slightest movement of the aircraft causes almost instantaneous reaction on the controls.
The heading is controlled by the directional gyro, in the nose of the aircraft, which operates both the pilot direction indicator (PDI) and the directional panel. Two wiper arms are on the directional panel, one each for ailerons and rudder, the latter pivoting on the directional panel. Ends of the pickups are attached to the dash pot links, preventing sudden skids from affecting the plane's heading by accelerating action of the rudder and aileron pickups. It then permits a gradual return to normal.
Using the turn control, the human pilot may make course corrections and even 360° turns without losing altitude. He also may bank up to 30°.
Current for the autopilot is supplied by the rotary inverter. The three servo motors get power from the aircraft's main supply. The motors move the controls to positions corresponding to that of the related gyro-operated wiper. They are locked in position until receiving new electric impulses. A three-part electronic amplifier (for ailerons, rudder and elevators) boosts the current.
Leveling the aircraft is done with a vertical flight gyro which causes the gyro case to turn when the plane pitches or rolls. This motion rotates the aileron control wiper, causing the servo motor to operate the ailerons or elevator. Pilots can bring damaged planes home with the M-H autopilot because the servo motors are placed as close as possible to the controls they operate. There is relatively little chance of enemy shell fire damaging the small, flexible wiring between amplifier and servo motors, while mechanical controls from the cockpit are in constant danger.
A Norden engineer explains the linking of the Norden bombsight with the autopilot:
"When control of the airplane is in the hands of the bombardier, the corrections and adjustments he puts into his bombsight automatically change the directive forces flashed by the gyros to the power unit of the autopilot. Thus the bomber is flown automatically by the bombsight . Changes are made instantly, with no lag such as human hesitation; there is no possibility of error due to the human equation."
Though it has limitations, such as inability to make violent evasive maneuvers and aerobatics, the M-H autopilot "can fly a plane with much more precision and accuracy than can a human pilot," according to Col R E Jarman, chief of the bombardment branch of the Wright Field Armament Laboratory, which helped perfect the device.
For example, the normal human requires one-tenth of a second to react to an outside stimulus. A bomber flies 50 feet in that time. Hence that tenth of a second can result in the bomb missing its objective by many yards, if wind, antiaircraft fire or rough air throws the plane off course at the last second.
But the autopilot reacts so swiftly to outside disturbances that a recovery is made almost before the disturbance has moved the plane from its course. Its sensitive mechanisms take control when flight accuracy is of the utmost importance. The pilot might have 10 different things to do mechanically all racing through his brain at the same time. He cannot do them all at once. But the electronic pilot can have 300 "thoughts" in its "mind" at a given instant and perform all of them within a minute.
Development of the device began under the direction of the Navy Department in 1935, with the C L Norden Company, builders of the Norden bombsight, making the first units. The Navy continued testing the device until 1938, when the Army took over the tests to adapt the pilot to its particular uses.
The first AAF plane to be equipped with the electronic pilot was the now obsolete Douglas Bolo. Convinced that precision bombing was to be the AAF strong point and that the automatic pilot (then known as Stabilized Bombing Approach Equipment SBAE) was the means of assuring precision bombing, a group of Wright Field experts set to work to perfect the device. Besides Colonel Jarmon, these men included Brigadier General Carroll, chief of the engineering division of the Materiel Command; Capt D E Hamilton and Col D N Kilpatrick, later killed in a crash.
The old SBAE called for special equipment on each type of plane. While basically what the researchers wanted, improvements still were necessary. Meanwhile, a group of Minneapolis-Honeywell engineers demonstrated a new electronic circuit at Wright Field. The original circuit was of no value as an autopilot, nor was it intended to fly an airplane. But technicians at Wright Field saw its possibilities and asked the company to design an electronic system around the then-existing Norden autopilot.
Minneapolis-Honeywell men and Army officers tested the equipment in Mitchells and Fortresses during the spring and summer of 1941. Late in the summer of 1941, Colonel Jarmon and another officer flew a Fortress from Wright Field to Washington to demonstrate the development to high officers of the Army and Navy.
Both services approved and the autopilot was put into production in October, 1941. Almost immediately it was standardized and installed in AAF bombers. The first standardized autopilot was installed on a Fortress early in 1942. It has undergone many modifications and improvements since then and today it is standard equipment on all US four-engined bombers and on some bomber-trainers.
Minneapolis-Honeywell first started by modifying the Norden-built servos, gyros and stabilizers at its main plant in Minneapolis. Later an entire new plant was equipped in Chicago for manufacturing the complete autopilot.
Production of the automatic pilot requires great precision and extraordinary special precautions. Minute particles of dust lodged in a critical part of the control will prevent proper functioning. Rust and other chemical actions, even in infinitesimal quantities, are major problems in manufacturing the autopilot.
Autopilots are assembled in glass-walled, air-sealed and air-conditioned rooms. Dust is filtered from the air by an electronic "Precipitron." Temperature and humidity of the room are controlled to prevent rust from perspiration and atmospheric humidity. Dimensional changes in metal caused by temperature variations are also minimized.
The Army Air Forces changed the Navy designation (SBAE) of the autopilot to AFCE (Aircraft Flight Control Equipment) and then finally to C-1, which is the model now standard on all AAF bombers.
A few of the changes from SBAE to the improved C-1 included adding an electronic follow-up system and replacement of the original mechanical system. Other minor changes also increased its sensitivity.
Proof of the amazing capabilities of the C-1 autopilot came some months ago from England where a Flying Fortress, its fuselage almost cut in two by the wing of a diving German plane, was flown back to England and landed by the autopilot. The regular control cables to the tail and rudder were unusable and the plane itself was so badly damaged that it collapsed of its own weight shortly after landing.
Even after the autopilot was perfected, the bombing problem was not solved. Only a few AAF officers were convinced of the device's merits. They were gradually won over, and trained crews of two or three officers and five or six enlisted men were sent into each war theater to assist and instruct pilots on operation of the device. Many times these instructors participated in combat missions to prove that "their baby" really increased bombing accuracy.
The 12th Air Force in Egypt and Lieut Gen George Kenney's 5th Air Force were the first to demonstrate the practicability of the autopilot. Now all bombers in all theaters are using it.
The autopilot is operated in the following manner: as soon as cruising speed and altitude have been reached the pilot sets the autopilot controls. While flying on the autopilot, the human pilot can turn the plane by operating small knobs and switches. The human pilot can take control directly in cases of emergency by manually exerting more power on the control surfaces than can the automatic system.
Upon reaching the point where the bombing run is to start, the pilot turns the plane over to the bombardier. Setting the bombsight and automatic pilot dials, the bombardier actually flies the plane, putting it into evasive action by turning the dials in front of him. After the bombs are loosed, the pilot again takes control of the plane and heads for home, using the autopilot until he is ready to land.
In the latest production models of both the two- and four-engined bombers, the navigator's station also is equipped with controls so that he, too, can fly the plane as well as the pilots and bombardier. The C-1 autopilot now is standard equipment on Boeing Flying Fortresses, Consolidated Liberators and the Beech Kansas trainer.
"On long flights, a pilot actually needs an automatic pilot to help him fly the plane without tiring," Colonel Jarmon points out. "Considerable physical pressure is required to operate the controls of these big planes."
Colonel Jarmon is sold on the autopilot. "We've never had a report that an autopilot was responsible for a crash," he states. "On the contrary, we've had about 50 instances where planes damaged in combat have been flown back to their bases by their autopilot alone because of damage to the plane's regular control mechanisms."
This article was originally published in the May, 1944 issue of Flying magazine, volume 34, number 5,. pp 48-49, 158, 160.
The PDF of this article includes diagrams showing the installation and operation of the C-1 autopilot, a photo of the autopilot control panel in a B-17, and a photo of B-17F 41-24406 in flight after being rammed in the aft fuselage by a German fighter.
Photos credited to Minneapolis-Honeywell, Bob Doty, AAF.