We admit it: The modern autopilot is a marvelous technological innovation, capable of smoothly flying even antique airplanes better than most of us on our best days. Especially when operating in the en route environment, its usually on, and doing its thing so we can concentrate on the scenery. Its also useful in other flight modes, of course, especially in the clag. Letting George do it frees our mental bandwidth for chart folding, communicating with ATC and planning the upcoming approach to minimums. Often, George will do a better job at flying that approach, but not always. Even so, he has to be monitored and the crew prepared to hand-fly the landing at a minimum, or intervene quickly if “something happens.” Many pilots like to fly autopilot-coupled instrument approaches as standard procedure. Like every other technology, however, autopilots have their limitations, especially when used close to the ground. Here are five reasons not to use the autopilot to fly an instrument approach. 1. Lower minimums A very frequent rationale for flying coupled instrument approaches is that “George” flies more precisely than any pilot. Consequently, the logic progresses, if the weather conditions are at or near minimums its better to let the autopilot fly the approach. A further corollary: Its safe to fly approaches to lower minima using the autopilot than you are comfortable with or capable of flying by hand. The first stipulation (greater precision) may in fact be true, but with a caveat. A big caveat. For any number of reasons the pilot (you) need to be able to step in at any moment and take over by hand. If you find yourself on a coupled approach below your personal hand-flying minimums and then something happens to the autopilot, will you be ready to immediately take over and fly with the precision needed to complete or miss the approach? If the answer is not a resounding “yes,” then youve disproved that its safer to use the autopilot to fly approaches to minimums. There are any number of things that can impede autopilot function or cause a disconnect at any time. The complete list of failure modes depends on the model of autopilot installed. The brand and models of navigation system (especially GPS) interfacing with the autopilot can introduce failure modes as well. A short list of possibilities includes electrical failure, attitude indicator failure, heading indicator failure, turn coordinator failure, accidentally or inadvertently touching the trim switch (in some models, bumping one side of a split trim switch will disconnect the autopilot but leave a flight director engaged, making it look like the autopilot is still doing all the work), trim servo failure, GPS or VOR/localizer system failure, antenna connection corrosion, pilot activation of the wrong autopilot or nav system mode, airframe ice accumulation (that impedes trim or control surface movement), turbulence, a right-seat passengers nudge on the control yoke or curious push of a button…the list goes on, limited by imagination and the vagaries of avionics installation and configuration. Are you ready to identify the failure and hand-fly the rest of the procedure, perhaps “partial-panel” or with degraded navigation? If not, youve let the autopilot fly you too low on the approach. 2. Proficiency Hang around a pilots lounge or Internet forums long enough and youll hear someone say they havent flown a lot of IFR lately, but the autopilot will take care of them. For all the reasons discussed above, this is a very bad plan. An autopilot is a very capable, very stupid copilot that will do everything you tell it to do and nothing you do not (whether you are right or wrong). But it has no ability to predict the future. In other words, its only as good as your ability to control it. If youre rusty “on the gauges” (or even more so because of its complexity, “on the glass”), chances are good youre going to be a little behind the curve on the autopilot and the systems driving it. This includes partial panel flight-because in many cases loss of the primary attitude and/or heading indicators will disconnect the autopilot, leaving you to go from monitoring a coupled approach to identifying a failed instrument and transitioning to partial panel hand-flying while still flying the procedure to a landing or a missed approach…a recipe for disaster if youre not very sharp flying by hand. Some autopilots make this easier for you. Although generally considered less precise, the typically lower-priced rate-based autopilots (S-Tec, for example) reference a turn coordinator, not the attitude and heading indicators, so they are still usable in the most dramatic partial-panel scenarios. More sophisticated attitude-based autopilots will bail on you the second the attitude gyro dies or a “Heading” flag drops on an HSI. 3. Complacency Along those same lines, you need to actively participate in a coupled approach. Some years ago I had a flight student who had a long career as a bush pilot in Alaska, but had retired to the Lower 48 and decided a turbocharged Bonanza was a better way to get around down here than the Super Cub hed flown Up North. Despite his superb stick-and-rudder skills I could not get him to safely hand-fly the airplane in the week I had available. My students response was that I had failed to teach him the “man-in-the-loop” concept of autopilot operation (a common term for this in the early 1990s). Trouble was, this pilot wasnt in the loop at all-when it came to IFR he was a “gear up, autopilot on” type who from that point on pretty much let the autopilot do its own thing until he was visual again at the arrival end of his flight. He paid very little attention to what was happening on the instrument panel between autopilot programming updates. And he absolutely could not hand-fly the simplest basic attitude climbs and turns by reference to instruments. “Thats what the autopilots for,” he told me. Pushing the buttons, then sitting back and watching for decision height and the runway environment wont cut it. You (and your passengers) are not just along for the ride; you are still pilot-in-command. Instead, do the same things youd do if hand-flying the airplane. Complete cockpit flow checks and checklists. Make callouts for altitude, heading and vertical speed. Adjust power and airplane configuration where and when needed. Otherwise youre setting yourself up for complacency. If you arent an active participant in the ongoing approach procedure, then you shouldnt be flying a coupled instrument approach. 4. Unfamiliarity If you arent intimately familiar with how the autopilot works, you have no business letting it fly an approach. Im guilt of impulsively pushing the wrong button at the wrong time, and I imagine youve done the same thing at least once. At another time, several years ago, I served as the pilot for a demonstration of the Garmin G1000 system in a then-experimental G36 Bonanza. Although the sky was severe clear and I was pilot-flying, my demo pilot was definitely pilot-in-command of the avionics. Even then we had one of the classic “Now what is the autopilot doing?” moments when trying to set up for vectors to a practice coupled ILS. If you ever find yourself asking that question, you know you need more time learning about the autopilot system. Take the time to read the autopilot supplement-some are quite extensive-and to learn the tricks of interfacing it with the model of GPS installed. The same GPS can do different things with different autopilots, especially if the GPS is an early model or the autopilot design is more than a decade or so old. Understand and perform the full autopilot preflight check from the supplement, and know the systems limitations (some may only be able to intercept a glideslope from below, for instance, while others are certified only when the fuel load is balanced within some parameter in one wing compared to the other). 5. The missed approach Missed approach procedures are predicated on climbing to a safe altitude, then turning in a safe direction to get away from terrain and other traffic. The autopilot/navigation system combination has no way of knowing how fast your airplane will climb in the missed approach, because that will vary dramatically based on airplane weight, environmental conditions and your piloting technique. Consequently, autopilots will not fly the missed approach; you need to disengage the autopilot and hand-fly at least long enough to get the airplane configured for a climb. This can involve power, landing gear, flaps, re-tuning navaids, GPS manipulation (into and out of “Suspend” or “OBS” modes) and communication with ATC. Thats a lot to do while hand-flying the airplane, so you need to be ready for it. Even if you break out and are in a position to land the airplane, you may have some manipulation to do. Further, autopilots have the capability of “holding a little pressure” against the trim servos, meaning its not unusual for the airplane to be a little out of trim when you click off the autopilot. Remember, too, according to the FAA, in most cases coupled non-precision approaches must be discontinued and flown manually at altitudes lower than 50 feet below the minimum descent altitude, and coupled precision approaches must be flown manually below 50 feet agl. Other Stuff Notice that except in the context of flying lower minima coupled than you would by hand Ive not said much about autopilot failures. The truth is most autopilot systems are very reliable, so operational problems are more often the result of operator error or failure of another system that affects autopilot operation. When they do go, however, autopilots tend to go dramatically, with trim movements until the point autopilot control force tolerances are reached, at which time the autopilot disengages with a radically out-of-trim airplane. Thats a handful! In some of the most recent, all-electric airplanes, the pitch trim also is electric. In fact, there may not be a manual trim system at all. With such a design, there is no way to manually trim off the extreme pressures…meaning youll have to complete the approach or fly the miss by hand while fighting the out-of-trim condition. Use your autopilot as the excellent cockpit aid that it is. But dont let it take you someplace you cannot comfortably hand-fly if you had to. Dont fly autopilot-coupled approaches for the wrong reasons. Tom Turner is a CFII-MEI who frequently writes and lectures on aviation safety.