Autopilots On My Brain!

by Michael J. “Mick” Kaufman
Published in Midwest Flyer – October/November 2020 issue

In recent weeks, I have been helping one of our Midwest Flyer Magazine readers in Brazil troubleshoot a chronic autopilot problem. I have seen many autopilot issues throughout the years, and some have never been solved totally. To claim that I am an autopilot expert is far from the truth, and when I have a tough question, I call my dear friend and colleague, Bill Hale of Ft. Collins, Colorado. Bill is an electrical engineer who designed computer chips for a living, and designed, built and got the autopilot on his Bonanza certified by the FAA. When I get a question about autopilots, I try to find out all I can by creating different scenarios, then try to diagnose the problem. If I cannot determine an easy solution, I call Bill. I can recall so many unique situations in my years of training that it would require a book to cover all of them.

I have been teaching flying for over 50 years, so I always feel that a history lesson is necessary to show how we have evolved to what we do today, and I feel it is necessary to do this with autopilots.

When I purchased my Bonanza 32 years ago, it had a Lear autopilot, which worked remarkably well considering it was state of the art in 1962 when the airplane was built. It could do everything that modern autopilots could do except no GPS functions, as GPS was not even a dream back in 1962. When I replaced this autopilot, I saved about 70 lbs as it was the size of a suitcase and used vacuum tubes.

When training pilots for their instrument rating prior to the invention of GPS navigators, autopilots needed to be designed to avoid errors in the navigation structure as much as possible. Going back to the Airman’s Information Manual (AIM), we learned and needed to know some of the shortcomings of the system, not only for the autopilot, but for the pilot flying as well.

The localizer and glideslope both produced false signals, and I would try to trick the instrument students into following one of those signals. Autopilots were designed as much as possible to avoid these false radio signals by having “timing circuits.” For example, the autopilot needed to see a localizer with needle alive for 20 seconds before the glideslope needle centered in order to capture and fly the glideslope. It was a rule back then, and I believe still practiced that ATC needed to confirm with the pilot if he/she would accept a turn inbound on an ILS approach if the turn would not allow the aircraft to be established on the localizer more than two miles from the final approach fix. This procedure would allow autopilot flown approaches to capture and track the glideslope due to the safety feature provided by the timer. Before the advent of GPS and our modern approaches, it was necessary for us to always verify our position by more than one source, such as a localizer and a VOR radial or a localizer and an NDB (LOM) or marker beacon. Sometimes we used DME to get an accurate position and verify that this was the correct localizer and not a false course. Today, when using the GPS to assist us in getting established on an ILS approach, it is very simple. We can be relatively certain we will not be fooled by false signals. Learn the characteristics of your autopilot and on that rare occasion that it does not do things as you expect, analyze why.

There are some bad features inherent in some avionics installations that could easily kill some unsuspecting pilot. Some of these installations require a specific combination of equipment, along with certain events during an approach.

One I remember from some 10-plus years back is an installation of a Garmin 430/530 in Bonanzas with altitude preselect and a King KFC 150 autopilot. This was a factory package option when the airplane was delivered as new and probably worked fine. When WAAS became available, most pilots decided to upgrade and sent the non-WAAS navigator in for the upgrade. When the navigator came back from the factory, every pilot wanted GPSS (GPS-Steering), so it was installed.

Now for the deadly scenario.

You are flying an ILS approach with either a DME arc or procedure turn course reversal. You are at your assigned altitude, one which would have you capture the ILS glideslope at the final approach fix. GPS steering is doing a fine job, you are at approach airspeed of about 120 kts and have a power setting of about 16 inches of manifold pressure. You are happy and study the approach chart for the last time prior to reaching the final approach fix. The autopilot begins its final turn inbound and the navigator is set to auto and goes to VLOC mode; the glideslope comes alive, and the needle, which was not previously visible, comes alive from the bottom of the course deviation indicator to display that the glidepath is now above you. As the glideslope transitions to above you, the autopilot sees this and uncouples the altitude hold and starts to climb to capture the glideslope. You are still in Instrument Meteorological Conditions (IMC) looking at the approach chart, as the airplane climbs and the airspeed decreases until the airplane stalls.

I know this is true because I tested this many times and brought this to the attention of the FAA and fellow instructors in our training program. I have seen about six different airplanes with this issue, and my instructor colleagues have also reported seeing this situation with this avionics combination.

As an instrument flight instructor very dedicated to pilot safety, I urge pilots to never fly in IMC after maintenance, software updates or equipment upgrades until you fully understand how everything is working together. Be ready to disconnect your autopilot, assess the situation, fly needle ball and airspeed, if necessary, and use the “E-word” (declare an emergency), if you need to.

Why does my autopilot usually capture the glideslope, but sometimes it doesn’t?

This is a very common question I hear.

The first test in troubleshooting is centered around that timing circuit that is still necessary in most autopilots. After that, we need to find out what always works and if possible, a scenario that never works.

We spent the better part of a day and $1,000 in fuel trying to troubleshoot a Garmin G-900X connected to a Tru-Trak autopilot in an Epic aircraft. The solution was somehow tied to the use of the “altitude preselect.” The choice for a successful glideslope was to either not use the altitude preselect or to disengage the autopilot after using it and then reengage the autopilot. We were never totally able to solve the problem, but we did find what worked every time.

A more recent example was in a Bonanza with an Avidyne IFD-540 and a King KFC 200 autopilot. In this situation, we found it necessary to disconnect the autopilot and turn it back on once the glideslope was alive on the indicator. On most WAAS navigators (except the Garmin GNS 480), the glideslope does not come alive on a GPS approach until the final approach fix is the next waypoint in sequence on the approach.

Many times, some of these autopilot problems can be traced to the pilot either not having a good understanding of his flight profile, or “just plane stupid,” when I was learning to use my newly installed equipment.

I was flying a practice localizer approach using GPS Steering on my #2 radio, which is the Garmin GNS 480 for GPS assist in getting established. Once inbound and 2 miles from the final approach fix, I switched the navigator to VLOC and the autopilot to approach mode with the localizer set to the proper frequency and properly identified. About 15 seconds after the switch, the airplane made a 90-degree turn, which I knew was not correct. I tried it again thinking I have a glitch and the same thing happened. What am I doing wrong? On the third try, I found the problem. My #1 Nav/Com is a King KX 175 and there is a toggle switch to tell the autopilot which radio Nav head to follow. I had the switch set to my #1 radio, and the radial dialed also on my #1 radio, by coincidence just prior to the final approach fix on the localizer. When this radial came alive, the autopilot in approach mode did what it was supposed to – capture the radial and fly it. I had always had this switch to select which indicator the autopilot was to fly, however in the process of using the Garmin GNS 480 which was new to the airplane, I forgot the switch still existed.

In conclusion, it can be said over and over again, know your airplane and equipment, and don’t takeoff in IMC after maintenance or equipment updates. I have even seen corrupt data base updates for navigators. Have a back-up plan, back up equipment, and know how to use it. FLY SAFE!

EDITOR’S NOTE: Michael J. “Mick” Kaufman is a Certified Instrument Flight Instructor (CFII) and the program manager of flight operations with the “Bonanza/Baron Pilot Training” organization. Kaufman conducts pilot clinics and specialized instruction throughout the U.S. in a variety of aircraft, which are equipped with a variety of avionics, although he is based in Lone Rock (KLNR) and Eagle River (KEGV), Wisconsin. Kaufman was named “FAA’s Safety Team Representative of the Year” for Wisconsin in 2008. Email questions to or call 817-988-0174.

DISCLAIMER: The information contained in this column is the expressed opinion of the author only, and readers are advised to seek the advice of their personal flight instructor and others, and refer to the Federal Aviation Regulations, FAA Aeronautical Information Manual and instructional materials before attempting any procedures discussed herein.

This entry was posted in Columns, Columns, Columns, Instrument Flight, October/November 2020 and tagged , , , , , , , , , , , , , , . Bookmark the permalink.

Leave a Reply