Attitude Indicators are essentially gyroscopes. Mechanical gyroscopes have a "rigidity in space" property, which allows them to preserve their orientation when the support rotates. However, like spinning tops, gyroscopes will also slowly right themselves up due to gravity; otherwise the AI will show pitch down as the plane travels forward.

If I keep an airplane in a balanced turn long enough, is it possible that the gyroscope will right itself to the fake gravity, and show that I'm straight and level?

What about ring laser gyroscopes?


2 Answers 2


A gyroscopic Attitude Indicator will not build up long-term error simply because of its erection mechanism in an extended coordinated turn. However, all gyroscopes will accumulate error over time, which must be corrected in some fashion.

A gyroscopic AI will display a turning error of 3-5 degrees in a 180 degree turn. However, this precession is reversed in the second 180 degrees of a turn, so after 360 degrees, the precession is eliminated and the AI is correct again.

To explain why this is, first consider the design of the attitude indicator: the pendulous vanes exert an erecting force when the gyro is not vertical with respect to downward acceleration. Gyroscope emphasizing pendulous vanes

(FAA AC 65-15A)

This causes the gyro to erect itself over time, and would in fact do so incorrectly if the aircraft were kept on a constant heading (if, for example, flown uncoordinated on a constant heading.)

However, in a coordinated turn, the case of the AI (including the pendulous vanes) rotates around the gyro. For the first half of the turn, the gyro is erecting in a particular direction. However, as the pendulous vanes move more than 180 degrees around the gyro, they add error opposite the original direction - removing the original error. Diagram of gyroscopic instrument in two positions

Note that while this fixes turning error in the long term, other sources of error are still in place. Precession caused by the fact that the gyroscope gimbals are not perfectly friction-free, which will vary in each individual instrument, will still happen - and the erection system will be unable to correct for it, as it's too busy causing and removing its own error.

A ring-laser gyro would be subject to the same problems, because it must still have a correction process of some sort. All gyroscopes, even perfect ones, are subject to apparent precession:

A freely spinning gyro tends to maintain its axis in a constant direction in space, a property known as rigidity in space or gyroscopic inertia. Thus, if the spin axis of a gyro were pointed toward a star, it would keep pointing at the star. Actually, the gyro does not move, but the earth moving beneath it gives it an apparent motion. This apparent motion is called apparent precession. The magnitude of apparent precession is dependent upon latitude. The horizontal component, drift, is equal to 15° per hour times the sine of the latitude, and the vertical component, topple, is equal to 15° per hour times the cosine of the latitude.

(FAA Flight Navigator Handbook, p. 3-7)

Any gyro, whether a ring-laser, MEMS, or gyroscopic, in a coordinated turn, will exhibit apparent precession and will have to be corrected over time if their only source of correction is acceleration.

  • $\begingroup$ Just found this page with great pictures of an AI from a spitfire. It's german, but you don't need to read the descriptions. However, I'm not yet convinced. The erection mechanism works slowly, it will correct the AI on a long flight, but not on a 360° turn. But does it really have no effect when flying a circle for hours? $\endgroup$
    – sweber
    May 12, 2015 at 7:30
  • $\begingroup$ Re "(if, for example, flown uncoordinated on a constant heading.)" -- seems to me that if the aircraft were flown uncoordinated on a constant heading, the erecting mechanism would align the gyro with the apparent "down" direction, which is the same as the true "down" direction. In other words a pendulum points toward the center of the earth, not toward the cockpit floor, any time that the aircraft heading is constant. Including in a constant-banked non-turning sideslip, or for that matter when the aircraft is started up on tilted on ground. Which is exactly what we want to happen. $\endgroup$ Oct 27, 2020 at 9:41
  • $\begingroup$ So the answer could be improved by modifying that one sentence. $\endgroup$ Oct 27, 2020 at 9:42

No, because in a turn the local “level” continuously changes.

Lets say you are heading north and turning right. So your floor is banked to the east and the gyro is accumulating bias to east too. But after you make 90°, you are heading to the east and your bank is now to the south, so the gyro is accumulating bias to the south while the bias to the east is already reducing. So the gyro will never show precisely correct level, but it will never indicate level in the turn.

Regarding whether ring laser gyroscope will behave the same, it is a bit more complicated. The righting of the gyroscope is not inherent property of the gyroscope. It is added to compensate for curvature and rotation of Earth. A gyroscope made not to right itself maintains the same axis of rotation relative to distant stars. But as you fly along the curved surface of Earth, and as the Earth rotates with you, different stars will be getting into zenith. So the gyroscope has to be continuously adjusted to the local level.

When you have single gyroscope for use as attitude indicator, it needs a righting mechanism whether it is a mechanical or ring laser one and that mechanism will cause the same kind (and similar magnitude) of error in turns.

However a complete inertial navigation system can improve the precision. Such system knows how you move and so it can calculate how the local level, relative to distant stars, differs from the initial and can include this correction. It will still do some righting, using average acceleration, to correct for drift, but it can take the average over longer time, so the error in turns grows more slowly.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .