Autopilot Troubleshooting

By James Brewer

Your guy on the inside at SEA

As an avionics tech, I have specialized in autopilots primarily. I find it quite fascinating using electronics to do motive work in flying the aircraft, and how interesting it is using the electromagnetic force to overcome gravity! At least, when it’s all working well.

When there is a problem with an autopilot, it helps to have an understanding of the system. All autopilot systems have the same basic tasks to perform and so have (mostly) the same type of components involved. In short, they need to know what the airplane is doing and what it is supposed to be doing. It does this through the use of sensors and feedback loops. (With feedback loops nested inside feedback loops and more feedback loops inside of those feedback loops.)

Let’s get into the basic components that an autopilot needs to function.

  • Gyro. Typically a spinning mass to let the autopilot (and pilot) know that attitude ofAuto pilot the aircraft. Although accelerometers are beginning to replace the spinning mass, they provide the same type of information to the autopilot. You could say it is the heart of the system, but in staying with the biology metaphor, it is more like the inner ear where we get our positional information from.
  • Computer. Something has to do all the thinking! They take on many forms, but like all computers they take in information and output control data. Newer ones are real computers with processor chips, but the older ones are able to process the data without changing it to 1s and 0s.  (personally, I find that quite amazing)
  • Control mechanism (servo). The muscle of the system. The part that changes electronics to motion. Yes. I know that technically the electronic and the mechanical energy both come from the electromagnetic force, so it’s still fundamentally the same thing being used, but it is a conversion none the less. It’s all the more fascinating when it’s a vacuum servo! I wish I could have been there the day those were invented! “Hey, why don’t we use this whoopee cushion to fly an airplane?” “Great idea! Let’s get started!”

That really sums up the basics for an autopilot system. Everything else is another layer added on to the system to change what the computer thinks it needs to do. If there is an autopilot problem, the basic system needs to be checked since all other functions are dependent on them.

In flight, the best way to check the basic system is to get the aircraft straight and level and engage the autopilot with no modes (or only flight director mode) in clean air. At this point the basic feedback loops are in use and under control by the autopilot. Any deviation will be an indicator of a problem. Such as:

  • A slow wing rock or a pitch porpoise can indicate a problem with the control mechanism; be that a drive motor, bladder, feedback motor, or loose control cables.
  • A fast wing rock or pitch porpoise indicate a problem with the gyro. This failure mode is faster because it is reacting to the feedback loop that is the airplane itself. Autopilot 2The gyro is putting out information saying that it is moving and the computer is commanding the control mechanisms to drive quickly to get back to where it is supposed to be.
  • Drifting or climbing / diving uncommanded. Just like the wing rock and pitch porpoise, this one has degrees of difference separating the failure modes. A quick turn or dive indicates a problem with the gyro, where as a slow change indicates a problem in the control mechanism.

If the autopilot system handles basic flight with no problems, add in some offset by changing the aircraft attitude. Or more simply, grab the wheel push it. Change the attitude and see how it responds. It should return to straight and level flight. If not, see the above points for suspected problems. If no problems exist, check the modifier buttons; UP and DOWN. These are generally part of the computer and change its internal settings. A failure here indicates a problem in the computer itself.

If everything is still checking out OK, you can move on to the additional layers of complexity and additional components (sensors). Start with Heading mode. If it does not follow the bug correctly, there is most like a problem in the component that provides the heading information (HSI, DG). Next move on to the Course/ NAV mode. If the system doesn’t follow the CRS pointer, then the most likely culprit is the component that has the course data (HSI/DG). If it does not follow the NAV needle it’s likely the NAV source (radio or GPS).

Staying with the roll axis, move on to Approach if the Course/NAV modes pass. If this fails but NAV was OK, it can likely be a faulty computer. There are gains in the feedback loops that are changed inside the computer when switching to approach mode. Move on to Back Course to check that the system knows the difference. Again, a failure here is likely the computer if all other tests have passed.

Moving to the pitch axis, engage Altitude Hold mode and make sure it holds. Manually change the altitude and see if the autopilot recovers correctly. Next, adjust the speed of the aircraft. This will cause the autopilot to adjust the attitude of the aircraft to compensate to maintain the correct altitude. Failures here indicate a problem with the autopilot system altitude sensor (generally not the same as an altimeter). Likewise if the autopilot is equipped with an Indicated Airspeed mode. The aircraft will adjust as necessary to maintain the indicated speed. Failures here indicate a problem with the airspeed sensor (most likely) and just as with the altitude sensor, it is probably not the same as the airspeed indicator.

Glideslope mode can be difficult to troubleshoot. But as with most inputs to the autopilot computer, it is likely the sensor (GS antenna and or NAV radio.)

Most autopilots go through quite a bit of Pitch Trim testing as part of their initial internal tests. And quite often it is manually tested on the ground as part of a preflight. Testing it in the air is the same as on the ground. Set the autopilot up for straight and level flight, and then offset the pitch axis with the control wheel. The auto trim should kick in to recover the aircraft. Just don’t offset it too much! (Not that I speak from experience with that. Many years ago. With reoccurring nightmares.) A failure here actually indicates the pitch servo as the main culprit as the sensors that tell the computer that it needs to adjust the trim setting are located in the pitch servo, not the pitch trim servo.

Aside from some other less common options, that sums up most checks and in general the additional modes are going to be related to their unique sensors. (½ Bank mode is just a nightmare so I don’t want to talk about it. Suffice to say, it’s probably the computer, or the servo, but maybe the gyro)

PanelThis of course is by no means a comprehensive troubleshooting guide, and comprises thousands of pages of technical information along with too much tribal knowledge compressed as densely and as concisely as practical. The purpose of this being to give an understanding of how an autopilot tech thinks about the system and provide a means of communicating a failure mode as precisely as possible to eliminate troubleshooting time and expenses. One of the good and bad things about autopilot failures is that they generally manifest themselves in flight. This is bad because it can be expensive to troubleshoot an odd problem, and good in that I enjoy test flights.

Of course all of the information above presupposes correct wiring and structural integrity. If there has been major work done to an airplane, check that area first before troubleshooting the autopilot. For example, if a wing has recently been replaced, it’s a good idea to pass that information on if there is a wing rock problem in the autopilot; before two techs spend 40 hours each troubleshooting, scratching their heads since it passes on the ground and fails in the air then finding out about the wing replacement and finding a loose connector that would have been looked at in hour one had that knowledge been available. Hypothetically of course.

Below is a quick checklist that I have used when checking out an autopilot system in flight. It’s quick and paints the problem with a broad brush, but it does help to narrow down complicated problems into manageable sections that can be addressed individually.

In flight (assuming self-tests pass)

  • straight and level
    • AP eng – level flight
    • Offset aircraft – recover
    • Maintain offset – trim function
    • UP / DN modifier
  • Roll tests
    • HDG mode
    • NAV
      • CRS pointer OK
      • NAV needle OK
    • APR – gains
    • BC – opposite gains
  • Pitch tests
    • ALT HOLD
      • offset aircraft / recover
    • GS – capture and follow

And as always, make sure to follow all applicable flight instructions, operating handbooks, pilot’s guides, and other such directives when flying.

P.S. I wanted to write about the recently discovered (maybe) 5th fundamental force that will (possibly) add to gravity, electromagnetism, the weak nuclear force, and the strong nuclear force. Unfortunately I couldn’t find a way to relate it to aircraft. Unless it’s the force that causes avionics to fail….hmmmm


Leave a Reply

Your email address will not be published. Required fields are marked *