Flight controls move in reverse when on autopilot! (roll axis)

Bug description

When applying manual control input with the autopilot engaged, the autopilot can be seen steering the in-sim yoke in opposite direction of your control input. The ailerons seem to be “connected” to your joystick while the autopilot moves a dumm.y control wheel in response. When giving too much input the autopilot disengages (as it should) and the in-sim yoke abruptly synchs with the joystick position.

Its almost comical, when moving the joystick left, the in-sim yoke can be seen moving right as if there are two separate sets of controls. It makes flying on autopilot using the keyboard almost impossible as the controls don’t “self-center” when engaging the autopilot. The autopilot is continuously fighting the manual control input…

See video below, in both cases I’m giving left control input, you can see the in-sim flight controls moving to the RIGHT to counteract and then disconnect as I’m applying more and more left input. At the moment the autopilot disconnects, the in-sim yoke suddenly synchs with the joystick position. On the pitch axis behavior is as before.

Once more on the TBM, giving steady roll input to the left, this time staying below autopilot disconnect threshold. The autopilot moves the in-sim yoke to the right and increases input as deviation increases. The autopilot pitch mode eventually over-stresses the airframe by pulling in an attempt to maintain altitude at a high bank angle which is also wrong / missing autopilot logic.

You can’t see my joystick, since this is very non-intuitive I want to stress again, my roll inputs in the above video are to the LEFT! In response, the autopilot rolls the yoke to the RIGHT! Also important to note that in the above video (TBM), I’m not increasing the roll input, I’m simply applying a small input on the roll axis and keeping it there.

How it should be

In real life you can’t physically move the controls with the autopilot servo’s engaged, when applying force on the controls with the autopilot engaged the controls won’t move, when applying sufficient force on the controls this will be sensed by the autopilot servos and this will trip the autopilot offline. They should make the autopilot completely IGNORE any manual flight control input below the autopilot disconnect threshold, if manual control input is excessive the AP should trip offline.

Also the autopilot should disconnect at stall warning activation or manual trim inputs!


So am I the only one or is this a bug? Its easy to test, just engage autopilot, move the joystick left / right and observe the in-sim yoke move in opposite direction.

Can confirm. Controls move and can be moved with autopilot engaged.
In the TBM with AP engaged, when control input is left/right, yoke direction is = opposite. Plane does move slightly in the direction of control input.
When pulling or pushing the yoke it moves in the same direction.


I filed it now as a bug, thanks! Weirdest bug ever though…

1 Like

I noticed the interesting yoke movement the first time on one of the iceland landing challenges, where the autopilot was engaged when starting it. Was very weird. This was before SU5 dropped.


Oh really, did it got unnoticed for so long? I’ve reinstalled MSFS after SU5 after being away for a long time, hadn’t noticed it before…

I don’t know if the behavior was exactly the same as in the video, but I can distinctly remember doing the landing challenge: Turned my yoke left, visually the yoke turned right (including the aircraft to a certain degree).
My first thought was that my control axis was reversed by the World Update, but then why does the aircraft behave so weirdly? So I actually looked at the instruments in detail and felt dumb when I saw that AP was engaged…

1 Like

Last time I flew (pre-SU5) was multiple updates ago, I was complaining that it was possible to fly the aircraft with the autopilot engaged, you could hardly notice the AP was engaged. Also when the dead zones were set too small the autopilot wasn’t able to overrule and the aircraft was in a continuous turn, climb, slip etc. until the joystick was completely centered. Now its completely opposite.

To make it realistic the manual flight control inputs need to be completely ignored until sufficient input is given to trip the autopilot offline. After which the controls should abruptly synch with the joystick position, same happens in real life when the autopilot all of a sudden disengages while applying force on the controls. I have never flown any aircraft were it is possible to fly the aircraft with the autopilot servo’s engaged, you physically can’t move the controls (apart from some play in the controls maybe) unless using Touch Control Steering (TCS) or Control Wheel Steering (CWS) but that’s the only exception.

The autopilot does feel this force you exert on the controls however as a out of trim condition and start to drive the trim as a result. I use this sometimes when I’m not sure the AP is trimming, pushing or pulling a little bit on the controls and observe the trim indicator. The controls still DON’T move in this situation, until exceeding the threshold for the autopilot to disengage.

The autopilot should also disengage at stall warning activation or manual trim activation, it still doesn’t.

1 Like

Back in the day (2007) when I’ve released FLT and FLN products (Flightline N - Reality XP Simulation) I was exactly doing this to flight test the ST55X Trim indicators and this has became a habit afterward whenever I was flying with this autopilot.

At one point in time I thought it is not right and I should stop otherwise once I’d get the chance to fly a real ST55X I’d be transferring bad habits from virtual to the cockpit. I feel less stupid reading you’re also doing it for real!

1 Like

I had received the Bravo on Tuesday along with the update. everything seemed okay.
Then the hotfix came while I was trying to configure settings with Spad.next. Then I started having issues
with AP not turning off, I spent hours trying to find the problem. Now when I engage the AP the yoke turns to the right and the trim is incontrollable. Turn AP off and it goes back to normal. Finally started FS2020 with all peripherals disconnected and engaging AP with the mouse I got the same behavior.
I’m so glad it’s not just me. It’s enough to drive a man crazy.

Maybe a dead-zone issue? If the joystick / yoke is not 100% centered, it is overruling the autopilot apparently. It is a bug of course, the sim should disregard the manual control inputs when autopilot is engaged, until enough input is given to trip the autopilot offline. The current implementation is flawed.

I don’t think that’s related to the topic of this thread; you might want to start a new thread to avoid confusion.

I have a Honeycomb Alpha and Bravo used via spad.next as well, and am having no problems with the autopilot activation or the disengagement when I press my AP disengage button.

Note that traditionally, most instances of “autopilot goes hard to one side when I engage it” are due to a key binding that is duplicated between something like the keyboard rudder control and something else you’re pressing, causing the keyboard’s “axis” position to be off. This is overridden by your actual yoke while it’s moving, but once input stops because you let go, the keyboard takes over.

I saw this a few weeks ago, with the Isafjordur landing challenge that was on the main menu (before Sim update 5).
The Caravan started with AP on, and it was all over the place.
Disengaging the AP was the solution for the challenge, but it was clear the AP had problems.

Yesterday I experienced the problem in a slightly different way:

I flew the TBM 930 for almost half an hour, most of the time with autopilot. After I pressed the wrong key during camera movement, I overrode the autopilot a short time. I think I had to reactivate the autopilot. The last thing I could remember was that the autopilot had to take a turn to the left, the yoke turned left but the TBM turned right! Worse: The autopilot steered even stronger to the left to a degree the simulation stated a crashed airplane!

Never experienced the autopilot steering in the wrong direction, steering more and more until it breaks the airplane.

Think this could the related to the reverse problem…

1 Like

Yes I have seen the same behavior, it happens when using the keyboard switches for flying instead of a joystick. Keyboard switches don’t auto-center like a joystick so it is continuously giving control input until moving the joystick slightly to center the controls.

Since I’m building my own force feedback yoke, I’m using various simulator variables via SimConnect library. Among variables concerning aircraft x-axis (roll) control, there are:
YOKE X INDICATOR - Percent control deflection left/right (for animation) (YXI)
AILERON POSITION - Percent aileron input left/right (AP)
When flying manually, both variables are equal and reflect my joystick deflection and yoke picture on the screen. Positive numbers mean deflection to the right.
But when the autopilot is on, strange things happen:
YXI is constant (let say 0) when joystick or keyboard are not touched, even though the yoke is slightly moving on the screen (especially in strong wind conditions).
AP is slightly changing (controled from autopilot). This movement is obviously reflected in the yoke picture on the screen.
So far so good. But when there is input from the joystick or the keyboard, let say to the right, YXI goes positive, AP goes negative (more or less the same value), and the yoke on the screen is moving to the left!
So the pilot is trying to move his yoke to the right, but the yoke is actually moving to the left.

New update, same same: