• E-Mail this Article
  • View Printable Article
  • Text size:

    • A
    • A
    • A

Nothing can be more exasperating to a pilot or aircraft owner than trying to get a balky autopilot fixed. But often more than half the battle is communicating the symptoms accurately to your avionics technician in language he'll understand. Autopilot wizard Gary Picou gives you an arsenal of autopilot terminology that you'll need to talk the talk.

On the Starship Enterprise, when something goes awry, an excited engineer-type says stuff like, "A warp-core breach is imminent!" This is accompanied by ominous humming and flashing warning lights to give you the idea that the ship is about to explode.

It never does. That's because the Captain knows to remind the engineer that in case of warp-core breach, the first thing to check is the magneto-resonance frequency in the dilithium crystal chamber. Too bad you can't similarly advise your avionics tech when your autopilot suffers the equivalent of a core breach. When discussing autopilots, your technician may utter words that sound like English but the only things that flash are green dollar signs.

We can't promise to tell you enough about your autopilot to fix it yourself, but we can explain the concepts you hear used in reference to the autopilot. And if you can describe its ills in terms your technician can understand, you could save some bucks in troubleshooting time.

No Simple Thing

Even the simplest autopilot is a complicated bit of machinery but all systems can be divided into three parts: A sensing part, a selecting/computing part and a control part.

The sensing part detects the airplane's status. Are we wings level? On the desired heading? Rolling in turbulence? The system has to know something about the current attitude and the forces acting on the airplane. These sensors are attitude or rate gyroscopes, pressure sensors, accelerometers, heading bugs, GPS receivers, glideslope pointers. Essentially anything that's an input to the autopilot computer.

The computing part looks at the information coming from the sensing part and compares it with the mode selection parts. The computer says to itself, "Okay, this is where we are now. Over there is where that human in the seat wants to go. I need to roll into a 22-degree bank to get there. Better haul away on the aileron cable."

So the computer commands a servo to move, yanking the controls to do its bidding. Those are the controlling bits, things like servo motors, clutches, capstans, cables and chains. (Look at a Mooney trim system, and you'd swear Orville and Wilbur took their bike shop to Kerrville.)

Flight control is a loop or series of loops. As the autopilot computer yanks the chains, it expects to see things happen. Some feedback is often sent from the servo, in the form of a position sensor or motor feedback. Plus, if all is going right, the airplane will go in the direction the computer's electronic brain wants it to go. If it doesn't—say because of turbulence—the computer applies more muscle until something happens.

It's the loop-iness of autopilots that gets the technician and pilot into trouble. On the ground, you have no loop. No matter how hard the autopilot tries, it just can't roll into a standard-rate turn on the ramp. Therefore, the best way to troubleshoot an autopilot is in its natural habitat, aloft.

A Little Self-Help

Plan on doing some diagnostics yourself. Otherwise, you risk big bucks by sending the repair person down the wrong path. The autopilot parts aren't something you can slide out with an Allen wrench in one minute for a bench check. The servos in an Aerostar with the added fuselage fuel tank are about 20 man-hours from daylight. That can be a $2000 gamble to change a roll servo, on a hunch. We'd rather go to Vegas!

Here, in alphabetical order, are some terms common to autopilot systems. Speaking the same language as your avionics technician is the biggest part of solving any problem.


Adapter boards, modules or sometimes "personality modules" configure the autopilot computer to a specific airplane type and are common in Bendix/King autopilots. Without the right adapters, the computer has a serious personality disorder, with unpleasant flight consequences. It's not unusual to find that a shop has stuck a loaner computer in, "Just to see what happens."

Because an A36 flies sort of like a Cessna 210, the real problem may appear to go away. Yet it's just buried. The adapter board in the KFC 200 contains all the aircraft alignment calibrations. Why? Because if you need to exchange the KC 295 computer, you can install the same adapter in the new unit and not have to re-align the system.

If the adapter board in the KFC 200 is replaced for any reason, or the attitude gyro goes bad, you should re-align the entire system. If not, you may introduce another problem or mask the existing one. There can be no mix and match of part numbers in any system. The correct equipment designed for that particular model airplane must be installed; no substitutions.

Following service, examine your logbooks and workorders. Verify that all part numbers, in and out, are recorded and that they match specs. If not, get an explanation.


Automatic electric trim or autotrim is controlled by the autopilot and is used to reduce pitch servo effort in the same way a human uses trim to reduce control wheel forces. Not all autopilots have autotrim, but all two-axis autopilots need trim. S-Tec and Century 2000 autopilots without autotrim incorporate trim "prompting." These are lights that remind Mr. Pilot to crank in some trim so the pitch servo doesn't have to work so hard.

Autotrim is a frequent cause of problems. You can have too much or too little. In flight, you should pay attention to the trim wheel. Does it run every few seconds or rarely? The frequency, even in a perfectly functional system, depends on many factors, not the least of which are the handling characteristics of the airplane. However, if the airplane has a pitch oscillation (see porpoise), you may have overactive trim, and mis-set trim sensors.

Trim sensors are not in the trim servo. They're associated with the pitch axis. They sense elevator loads, either from inside the pitch servo, or on the elevator cable. Exactly where depends on the brand and type of autopilot and airplane. The lesson is: If the trim is too active, don't replace the trim servo! If the autopilot technician recommends it, get a second opinion.


Analog autopilots contain lots of transistors and resistors and plenty of integrated circuits, but their stimulus and response are not controlled by a microprocessor. Century, S-Tec and some Bendix/King systems such as the KFC 200 are analog. The Bendix/King KFC 150 and later autopilots, as well as current Honeywell and Collins systems are digital.

Analog systems are easier to troubleshoot. Sometimes. But they have more components and therefore are more failure-prone. Airplanes are analog machines and don't lend themselves to digitalization easily. But so is music.

With enough computational power, anything is possible.

Attitude Based

Attitude-based autopilots such as the Bendix/King and Century systems (except for the Century I) use a sensor that detects the aircraft position from the attitude gyro. They manipulate the controls to put the airplane in an attitude that will satisfy the pilot-requested input for pitch (usually altitude) and roll, (heading). Attitude-based autopilots also use the rate of change of the attitude to control the airplane smoothly.


A single-axis autopilot has aileron control only: typically wings level, heading hold or heading select and some sort of coupling to a nav. Single-axis autopilots from S-Tec and the Century 2000 can be upgraded as your needs and desires change. Bendix/King autopilots are installed from the git-go as single axis. By the time you change all the components necessary to upgrade the system, you might as well have bought a new system.

Two axis autopilots have aileron and elevator control, possibly including pitch trim. You set altitude hold and pitch commands, such as an attitude or rate hold. Most will include glideslope coupling. It's important to understand how the glideslope mode works, what the prerequisites are and if the beam can be captured from above and below. You can find this stuff in your AFMS or approved flight manual supplement. To be perfectly ramp-check legal, you must have a current supplement aboard.

A three-axis autopilot includes a rudder servo. Most are yaw-damping systems that keep the tail where it belongs. Some airplanes—Bonanzas, for example—need yaw damping more than others.


A slip clutch allows the human to overpower the autopilot. Clutches represent a compromise between performance and safety. Clutches are strength limiters and therefore degrade the ability of the autopilot to efficiently fly the airplane. But in order to protect the occupants in the event of an autopilot malfunction, manufacturers install slip clutches to give you some control.

It won't be a picnic to fight a clutch, but you ought to be able to arrest the undesired control movement while disconnecting the system. The slip clutches are sized during certification to balance the malfunction controllability with performance. For this reason, the tolerance of the clutch is important. If the clutch torque is too low, the autopilot will be sloppy. Too high, and you won't be able to wrestle the airplane away from a deranged autopilot.

Will you be able to detect a clutch set too tight? Probably not. We recommend that autopilot clutches be checked annually. It will cost money, so you might want to wait until something goes wrong to see if you can control the airplane. Again, we confine our gambling to Las Vegas (or maybe Atlantic City, which is where the FAA does its gambling).

Some Century (a.k.a. Mitchell, Edo-Aire or Piper) systems use a shear pin, which has no sense of humor. Avionics shops love 'em. One slip and the autopilot is in for service.

Digital Systems

Digital autopilots employ a microprocessor to fly the airplane. They have fewer parts, because the microprocessors take over for many discrete circuits. The problem is most avionics technicians are analog devices. They have an inherent mistrust of computers (some even mistrust transistors). The troubleshooting process for microprocessor-based autopilots is different than doe an analog unit. You have to think like an autopilot.

How a digital autopilot flies is like this: The sensors receive the analog position (attitude, rate, CDI deviation, etc.) and input that to the autopilot computer. An analog to digital conversion takes place inside the box. The microprocessor compares the input with the desired result and outputs a data stream to command the controls to move. A digital-to-analog converter on the output changes the data stream to analog voltages to spin the motors and yank on the controls. Got it?

One word of advice, though. If you think the microprocessor is bad, or has a software problem, think again. Barring lightning strikes, the processors are the most rugged things in the box.

Manual Electric Trim

Manual electric trim is controlled by the pilot, an electrically driven trim wheel. Not much to say except that the switches wear out and more frequently, the wires inside the yoke get broken. In any case, manual trim problems tend to be simple and annoying, if not expensive to fix.

The best wiring is a coiled cord from the wheel, but for some reason (probably aesthetics), airplane manufacturers keep shoving this mass of wires through a tiny tube. It's important to describe trim problems accurately. If you grab the wheel or crank and trim the airplane, this is manual trim. If you run the trim with a little button on the control wheel, it's manual electric trim. If the autopilot trims, while you get 40 winks, that's autotrim.


A wise autopilot technician once said, "Fish porpoise, airplanes oscillate." Okay, there are purists everywhere, but after a couple of, ahem, oscillations, you will say, "Whoa, what a porpoise!"

What you are experiencing is an inability of the autopilot to maintain a stable flight path. There are two root causes. Too much control and too little control. It's your ability to classify these gyrations in flight, between onset and the time you disconnect the autopilot (which you must do for a safe conclusion of the flight), that will determine the relative success of the technician's troubleshooting.

The best cue is the deviation from the desired path. Under-controlling "fish" deviate from altitude a long way and over a period of some minutes. You can watch the controls as the autopilot drifts off farther and farther, then catches itself and wanders back on track. Typically, it will overshoot and go off in the other direction. The trim may run, but tend to be too little too late. Trim may aggravate the situation. What's more important, trim will be behind the autopilot. This kind of problem is typically a slipping clutch, miss-aligned or defective attitude sensor or even a dead pitch servo.

At the other extreme, an overactive autopilot will stick close to the desired flight path, but will be like riding a cantering horse. Avionics technicians look forward to these, because there's usually loose change in the seat cushions. Rapid oscillations are often the result of mis-alignment (too much gain), or a failure in the smoothing portion of the autopilot loop. Overactive trim will induce a porpoise as the pitch servo acts to counteract a momentarily out-of-trim condition. Pitch porpoises that inhabit the roll axis are often termed "wing rock" or "Dutch roll." Again, amplitude and period of the phenomenon are critical bits of information. Get your watch out and time it. Describe the amplitude in terms of degrees of roll.

Rate Based

Rate-based autopilots such as the S-TEC, Century I and Brittain systems depend on acceleration, movement along the longitudinal axis and altitude changes sensed by a transducer to determine attitude changes. They essentially ignore aircraft position, controlling the roll and pitch to put the airplane in a steady state while satisfying the command for heading or altitude.

The advantages are an independence from the attitude gyros, which have rarely set records for reliability, particularly when the reliability of the vacuum system as a whole is factored in. They give a true standard rate turn, instead of the approximation that a position-based system can provide. In addition, an attitude-based system can, in severe turbulence, attempt to bend the airplane trying to maintain an attitude or altitude.

In the S-Tec system, based on rate, motion extremes are sensed and overcontrolling is thus avoided. The biggest advantage is simplicity. The biggest disadvantage is precision in putting the airplane in the correct attitude. In 90 percent of general aviation airplanes, in 90 percent of flying situations, you'll never notice the difference.

If you know your autopilot system well enough to describe it, you'll be surprised how quickly a technician might be able to zero in on your problems. Communicate the facts carefully, make note of your observations and pass them along. Worst case, you'll at least understand what he's talking about when he says you've got an autotrim problem. If possible, take your shop guy for a ride before pulling panels and boxes.

But until the autopilot is working 100 percent, do the right thing: Keep it turned off and breaker pulled until you know what's wrong.