In researching General Aviation autopilots it seems to me that most commonly the way control surfaces are affected is via a servo that attaches (via a clutch) to the control cable. References here and here.

Are there any autopilots that have a servo that directly attaches to, and actuates the control surface(s) as opposed to indirectly via the control cable?

  • $\begingroup$ Related: How did old autopilots work? $\endgroup$
    – Manu H
    Commented Feb 7, 2020 at 7:47
  • 2
    $\begingroup$ The pilot is supposed to be able to "overpower" the autopilot causing a disconnect. Once the autopilot disconnects, the servos shouldn't provide any input to the control surfaces. This is a lot easier to do when you run the cables around the servo and use a clutch rather than directly connecting to the control surface. $\endgroup$
    – Ron Beyer
    Commented Feb 7, 2020 at 13:05
  • $\begingroup$ Sounds like an answer @RonBeyer. Of course, it doesn't answer "all", but it's a good indicator. $\endgroup$
    – FreeMan
    Commented Feb 7, 2020 at 18:10

2 Answers 2


From the Approval of Flight Guidance Systems AC-25-1239-1C document:

30. Override of the FGS.

a. General. An override of an engaged FGS function is defined as an action taken by the flightcrew intended to prevent, oppose, or alter an operation being conducted by the FGS function without first disengaging that function.

b. Autopilot override.

   (1) With automatic disengagement.
       (a) Override force. The autopilot should disengage when the flightcrew applies a significant override force to the controls. The applicant should interpret “significant” as a force that is consistent with an intention to overpower the autopilot by either or both pilots. The autopilot should not disengage by minor application of force to the controls, such as a pilot gently bumping the control column while entering or exiting a pilot seat during cruise.

(Page 37)

It goes on to say that 25lbs of force has been determined to be a significant override force level for "other than approach" operations. This may be difficult to measure using a system that just looks at the force on the control surface since normal operations may cause force feedback through the surface. There would have to be a way to measure the pilots force in conjunction with the control surface force and determine if the pilot is trying to override the servo.

GA aircraft are not normally equipped with all the "bells and whistles", autopilots are typically optional equipment. This means that the autopilot system needs to "fit in" to the existing control system architecture. Adding bosses, actuators, connection points, holes, etc to an airframe to accommodate an autopilot is a "significant modification" that may require a new type certificate or an STC.

The easiest way around this is to use servo controls with clutches that move existing control cables. This way the pilots force is directly measurable on the control system. Using magnetic clutches the autopilot can easily disconnect itself, worst case, by popping the autopilot breaker (although this is not an acceptable disconnection procedure).

I can't say that there are no GA aircraft that don't use a different mechanism, but I've never heard of one.


To be able to drive a primary control circuit, the AP uses what is in effect a bi-directional cable winch that allows a small electric motor to apply the normal forces a pilot does. Pretty much all conventional cable operated flight control systems use that kind of A/P servo. So you can say that the A/P is driving the control surface directly, just like the pilot, but the A/P servo is a fairly "weak" pilot in that it's maximum force is less than a human pilot can generate, only what is needed for normal control.

The servo's electric motor drives a winch capstan drum through a reduction gearbox, and the winch cables are tied in somehow to the cable circuit itself to be able to pull on either side of the cable circuit, or into a bellcrank in the circuit when the motor is energized in one direction or the other.

There are two clutches between the motor and the capstan drum; an on-off tooth clutch and a slip clutch. The tooth clutch is two discs with radial gear teeth around the perimeter that intermesh when the discs are brought together. This clutch is operated by the A/P computer when the A/P is engaged/disengaged by the pilot (or disengaged by the A/P computer for various reasons), and is either fully engaged or not engaged. They are normally spring loaded disengaged and have to be energized by solenoid to be engaged, so they are normally fail-passive electrically speaking.

But just in case they do jam together for some reason, just upstream or downstream of the tooth clutch is a passive slip clutch that is there in case the tooth clutch can't disengage, so that the control circuit is not completely immobilized by a dead servo that can't disconnect (it also limits the force the servo can apply).

You will be able to move the controls by forcing the clutch to slip, but the effort is very high and it is quite unpleasant to have to control the airplane with controls that feel like they are set in partially cured concrete, but at least you can move them.


You must log in to answer this question.

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