Practical IFR: Don’t Disable. Revert!
|
Getting your Trinity Audio player ready...
|
They say automation breeds bad habits, but I think automation training is where the blame lies.
Here’s one beef: What should you do when the autopilot fails to capture the glideslope or turns right when you expected left?
You should disengage the autopilot and hand-fly, right?
Wrong. Wrong. Wrong.
Think about this objectively for a moment. Right at a critical moment in the approach, you’ve been hit with a surprise, so you double your workload by throwing out one of your best IFR tools. You do it right when a precise flying action is required. And you’re distracted because part of your attention is off thinking, “Why did the autopilot do that?”
But Disengaging is Easier
Yes, disengaging the autopilot is the “easiest” way to fix the situation, and that’s the problem.
It’s easier because we rarely do what I’ll call “Reversion Training.”
With one exception (which we’ll talk about in a moment), I have yet to see an autopilot surprise anybody in heading and basic altitude mode. (If it does, the thing is probably broken, and then we’re in agreement it should be turned off.) The reason is simply that heading mode and simple altitude or vertical speed hold are direct commands for performance. Fly left. Go down. Stay here.
This means that even in a critical moment, using these simplified functions should be an easy way to command your aircraft without abandoning the autopilot assistance altogether. When the more complex navigation or approach modes let you down (usually because something was entered incorrectly or too late), revert to the simpler heading and altitude modes and put the airplane where you want it. You know where the aircraft should go, or you wouldn’t be complaining the autopilot is misbehaving.
Building this habit of reverting down one level of automation takes a little practice. We have to break the red-button-disconnect habit, and we must build some skill using heading mode for more than vectors.
Step one is probably changing how you engage your autopilot. Do you go straight from hand flying to NAV mode where George follows the pretty magenta line? Don’t.
Instead, start with a Roll-and-Pitch hold mode if you have it. Essentially that’s the simplest autopilot mode of all and it’s easy to see if it’s working. When you engage roll and pitch holds and release the yoke, nothing should change.
The next step up is Heading Mode with a selected Vertical Speed or Altitude Hold (or just trim if you have only one axis of AP control). How’s that working? Great, basic autopilot modes are working. Now you can take the last step up to advanced navigation modes like GPSS, NAV, and VNAV.
Making a practice of stepping up lays the foundation for stepping back down when you need to. Part two is practicing entire approaches using just the heading bug and basic vertical speed control. It’s not hard. In fact, it’s kind of fun, but it takes some practice. You should be completely comfortable flying both ILS and LPV approaches with a continuous descent and non-precision approaches with level-offs and power changes using HDG, VS and/or ALT, and the throttle(s). Yes, you must also be comfortable hand flying approaches in case the autopilot completely fails, but that’s a different article.
There’s one other habit that’s useful for many reasons, but essential here. Make a habit of syncing your heading bug to your current heading on a regular basis, even if the bug isn’t in use.
The one time reverting catches people off guard is when they engage the autopilot HDG mode without realizing the heading bug is 110 degrees to the left. The aircraft dutifully rolls off toward the bug as the pilot makes a mad scramble to the swing the bug back forward.
To successfully and smoothly revert, you must have these details covered. Master that, and a misbehaving autopilot is almost boring.
Watch This Video:
“How to Fly a Flight Director”
Hand-Flying the Easy Parts
The simple fact is you get good at hand-flying an aircraft by … hand-flying an aircraft.
Something you notice watching many pilots fly is that few have trouble hand-flying when they’re focused on the gauges. The actual motor skill is not the weak part. The weakness is in split attention.
Out-of-practice pilots get into trouble because they’ve lost the skill of maintaining a pervasive and constant awareness of the flight instruments while they do other tasks.
Here’s a good exercise for that: Don’t use the autopilot in cruise. Use it for climbs or descents as you get ready for approaches if you want. Use it for at least some approaches as well. But when things get boring, turn it off. Your mind will naturally wander—forcing you to practice continually checking back to the flight instruments. Regularly reinforce hand-flying skill when life is relaxed, and it’ll be there for you when things get busy.
For 19 years, Jeff ran many noteworthy aviation media projects with his own firm, Van West Communications, including magazines, books, videos and live seminars. Jeff previously served as editor-in-chief of IFR Magazine and co-editor of Aviation Consumer, and his work appears in AOPA Pilot, Flight Training Magazine, Plane and Pilot, and AVweb. He’s an experienced CFII/MEI with ratings for single- and multi-engine airplanes, seaplanes, and gliders. Jeff was the creator of the first pilot transition program for new Cirrus aircraft.
- Practical IFR: Off-Route Thinking - November 18, 2025
- Practical IFR: Don’t Disable. Revert! - October 14, 2025
- Our Faith in NEXRAD - September 16, 2025
Good tips! I must admit that I am guilty of hitting the red button when things go awry, but I will work on re-training myself to fall back to HDG & ALT modes. For my part, this happens most frequently when I incorrectly activate an approach on my Garmin GTN, which inevitably attempts to navigate back to the start of the approach.
It’s easy to forget the sequence of button-pushing as a casual flier, so I would love to see a “quick reference” article for flying autopilot-coupled approaches using Garmin equipment. It all comes back to me once I’ve done it a few times, but it’s easy to forget after stepping away for a couple of months.
Common flubs for me:
– When to activate the approach in the GTN
– When to activate a segment
– When to activate Approach mode on the autopilot
– Forgetting to lower the altitude bug when using VNAV to descent through the approach
I was asked once by a close CFI friend of mine, “What do you recommend as a good way to do a GFC500 checkout?” My answer immediately was: “Have them fly the flight director only.” He just looked at me and said, “Holy cow! That’s a great idea!”
By flying the flight director only, you really start to get a deeper understanding of your autopilot. Remember, the A/P button is just enabling the servos to receive commands from the F/D! By just following the command arrows religiously it trains you in a few ways: 1) It forces you to hand fly which is always good and 2) When you screw up, you need to get the arrows to correct the screw-up but without the distraction of the plane really changing direction (so now you have more bandwidth to fix the screw up instead of jumping for the big red button) and 3) when you get it right with the A/P then on during a normal operation, you have greater situational awareness to catch when the automation goes wrong for mechnical/software reasons.
I’m a big fan of teaching flight using the FD command bars without the AP engaged.
Love it!
Jeff, great article. Syncing the heading bug frequently is a MUST. In the 777 I fly, the company has beat that into us as a habit, and it has saved me many times when told to fly a heading and not have the airplane beaming off left and right for no reason. In my GA planes that habit carries over. Great habit to get into if you don’t do it.
Jeff: Good article. Another habit I have adopted is to line up the VOR display course with the GPS course, even though it doesn’t “do” anything. What this does do is serve as a ready reminder of my heading, and (at least on Garmin 430/530 units) when the OBS button is pressed, the GPS magenta line skews with changes in the course selected on the VOR head. Very handy for using the GPS to set up the inbound course on a holding procedure.
Jeff, we have disagreed previously. many years ago. I suggested that cross-checking standby instruments was important and you disagreed in bold print. At the time I needed to be polite and left it alone, As I began reading this article I knew you were the author.
In the matter of not disabling the A/P when things are not going well as a general rule I disagree. inside the final approach fix as the PIC and observing pilot, if the flying pilot attempted to resolve an A/P excursion by pushing buttons on the mode control panel, I would push the red button to disable the A/P and state “my controls” without hesitation. The problem is that there is a malfunction or the FD has been miss-programed. It’s time to hand fly or hit the GA button. The time to correct either situation is when a single pilot or both crew members can safely determine the cause of the problem.
I’m always up for a respectful discussion with differing viewpoints. Thanks for weighing in. My comments here are specifically for single-pilot IFR. If it’s a crew environment, that’s different. The AP can be disconnected and the PF can focus fully on the flying.
With SPIFR, there are half the brain cycles to handle the situation, so reversion allows automation to still hold some of the tasks. I agree that the system is most likely behaving as commanded—but the command entered by the pilot was in error. However, my observation time and again is that virtually no one misprograms HDG and ALT or VS. So the reversion has little risk of operator error.
If the AP is still behaving oddly, then full disconnect is the right move. But if the behavior is as expected in these simpler modes, the pilot has more mental bandwidth to assess the situation. Say the issue was the programming of the GPS itself and the approach should be aborted because the guidance is incorrect. That’s much easier and quicker to discover when you’re not also handflying in IMC.
I will be the second to disagree with the article. Think about it objectively for a moment and from a cognitive processing perspective.
An autopilot is a layer of abstraction with regard to aircraft control. If the autopilot has done something unexpected, especially at a critical moment, it may not have been setup correctly, or it may have actually misbehaved. It is a cognitively more complex task to re-program the autopilot to address the airplane not flying the proper path than it is to manipulate the flight controls. Changing modes, even to a simpler mode is still reprogramming the autopilot.
The pilot in the moment is aware the that the airplane has not done what the pilot intended. It is in those moments the pilot needs to command the airplane, not diagnose or reprogram the autopilot.
When it comes to the adage, aviate, navigate, communicate, the autopilot fall squarely in the navigate category. When something unexpected happens, it’s time to focus on aviate.
“When it comes to the adage, aviate, navigate, communicate, the autopilot fall squarely in the navigate category. When something unexpected happens, it’s time to focus on aviate.”
I challenge that classification. Take the example of right after takeoff into IMC. You have established a steady climb by hand—aviate. You now engage the AP in a HDG and IAS or VS mode to maintain that steady climb. Is that navigate? I say that’s still aviate. So, engaging the same combination to put the plane into a steady flight state while you *then* attend to ensuring there isn’t a navigation problem is doing the equivalent of what we do stepping up in complexity: Handfly, basic AP, advanced AP.
I also think we drastically overestimate the average pilot’s ability to deal with the unexpected. Taking over the airplane by hand is rarely an issue. But then what? The other tasks of navigate and communicate must be attended to, and doing them while suddenly handflying makes those tasks much harder—while also risking loss of control from inattention or inability to aviate.
The cognitive problem I have seen students have with the autopilot is that when it does something unexpected, they usually freeze and ask the question, (often out loud), “why is it doing that?” Confusion has set in about the very thing you are asking them to reprogram. Like the other poster above who disagreed, I instruct in jets. Things can deteriorate very quickly unless positive control of the aircraft is maintained when the unexpected happens. Programming the autopilot out of the situation rather than taking positive control of the flight controls can quickly lead to a pilot deviation or worse.
I think one could critique you bring up (and I actually think Jeff may agree with you) is that the “rate of change” from a known state to an unknown one can be the deciding factor: In almost all single-engine pistons, for instance, you have far more time than in a jet to make A/P corrections. My intuition would say that rapid state change yields greater pilot reaction – for better or worse.