Systems

In parallel with Elmer Sperry's marine Gyropilot, Lawrence Sperry started to develop a fully-fledged autopilot for aircraft. Tragically, he was killed in an air accident in the English Channel in 1923. His younger brother, Elmer Sperry Jr., pressed on with development, and the Sperry A-1 Gyropilot was launched in 1931.

By the end of the 1930s, autopilots had become common in commercial and military aircraft, but as aircraft became faster, accuracy, lag and damping needed to improve even more.

In theory, electrical sensors should have given the fastest readings, but any method of measuring rotation interfered with the gyrocompass to some extent. Even electro-magnetic sensors, working like miniature generators, needed to produce so much current to control something like an autopilot, that they put too much load on the gyro-gimbals.

Work on autopilots continued around the World. In Britain, the Royal Aircraft Establishment began to develop their own autopilot in 1925, under F. W. Meredith and P. A. Cooke. Their objectives were to provide the R.A.F. with a stable bombing platform, and to investigate the feasibility of unmanned flying bombs. They decided that an all-pneumatic system would give a faster response than existing electro-hydraulic arrangements, which was true, except that the smallest amounts of grit tended to jam the pneumatic valves.

In Russia, Nicholas Minorsky conducted experiments to see how a helmsman keeps a ship on a constant heading. He observed that they anticipated which direction they were turning by sensing how quickly the turn was happening, immediately applying an opposite correction at precisely the same rate. It’s how we balance: corrections are often minute, without stopping to measure how far we’ve toppled; it’s self-damping, and we can scale it instinctively to be in sympathy with the inertia of the human body, or a smaller object we’re holding, or thousands of tons of ship. This was the rate/rate principle.

Minorsky emigrated to America in 1918, and

continued his work by joining the U.S. Navy. Sea trials were conducted in 1923 with some success, but not enough to persuade the Navy to spend any more money. Technology was not up to the invention, the project was shelved, and the patents were sold to Bendix.

Similarly in Germany, where Askania, Patin and Siemens joined forces to produce the world’s first all-electric autopilot in 1932. It was extremely sophisticated, working on all three axes, yaw, pitch and roll, and used Minorsky’s rate/rate principal. Interestingly, Siemens was also the first to register the name “autopilot”. Again, the prototypes were not as successful as they hoped, limited by contemporary electrical technology, and the project failed to win any further funding.

Despite teething problems at the R.A.E., reliable pneumatic pick-offs evolved, and became the most effective pre-war solution. For example: the Sperry A-2, a two-axis autopilot with pneumatic pick-offs and electro-hydraulic servos. Sperry also had a plant in England, and the A-2 was specified by Imperial Airways for the de Havilland DH-91 Albatross.

De Havilland DH-91 Albatross cockpit, with Sperry Autopilot control panel (No. 6)

FLIGHT