Should/does GA Autopilot coordinate turns?

[aurel42 10]

Hi there,

 

I have an obscure problem that hits me from time to time, I have no idea what causes it and would be grateful for any advice.

 

In my experience, the normal behaviour of autopilots in FSX/P3D is that they do clean, coordinated turns. 

 

In my current P3D installation (and I've experienced this before on a separate installation), the autopilot forgot how to coordinate turns, so it starts banking, the ball goes out the center, after a few seconds the ball (and the aircraft) swings violently in the other direction (is that a "dutch roll"?).  In effect, I have to assist the AP during every turn by coordinating with the rudder pedals.

 

So here are my two (groups of) questions:

1. How do Autopilots in GA aircraft work in the real world? Do they have rudder authority even when they don't have a Yaw Damper feature? Are they limited to bank ankles that don't cause that slipping behaviour?
2. Any idea what "overall" settings (ie. non-aircraft specific settings, I did not touch any aircraft config files) could modify the AP turning behaviour in P3D, what I could've messed up to make P3D act this way, and perhaps even how to fix it?

Additional info: I don't use "auto rudder" as I have rudder pedals. Since the behaviour changed, I've observed it, for example, in the Milviz C310R, RealAir Legacy and Turbine Duke.

 

Am I just being stupid somehow?

 

Cheers,

Marc

 

PS: If my explanation is unclear, I can provide video of the behaviour.

 

[JYW 2,965]

For the more general point, it just depends on whether the GA autopilot is 2 axis (eg. KAP140) or 3 axis (eg KFC225 w/yaw damper). A 3 axis unit will co-ordinate turns - a 2 axis unit will not, although this is rarely an issue in reasonably calm winds and at cruise speeds.

 

What you're seeing in the sim sounds quite extreme though - I've never seen dutch rolls occurring due to an unco-ordinated AP turn. You could check the aircraft.cfg for an aircraft that this is happening with; check the [autopilot] section, and ensure the yaw damper gain entry is set to 1.000.    Then, in case the aircraft does not have a dedicated yaw damper button, use the standard FSX assignment (CTRL+D by default) to engage the YD, and see if turns are then co-ordinated.

[aurel42 10]

Hey Bill, thanks for your input and advice.

 

I guess that means I've got something else going on that affects the behaviour of aircraft during turns and seems to make a lot more rudder action necessary — I'm seeing the same extreme yaw tendencies when the aircraft is mine, and it feels like I need quite a bit more rudder than before to do coordinated turns.

 

But, yeah, using the YD definitely helps, although I never had to use it before in the aircraft I listed. And I didn't touch the aircraft.cfg files for those aircraft before the issue started.

 

I shall try to identify whether it's some addon that interferes. I recently started using "FsPassengers" which does seem to change some sim settings (especially realism settings) when activated, so I guess that would make it a possible culprit. I believe the problem is much more pronounced on right hand turns — if it is in fact an asymmetric effect, perhaps it's something crazy like P-factor being set to more than 100% (if that's even possible).

[Gregg_Seipp 565]

Just a thought...have you calibrated your controls?  The reason I'm asking is because I've never seen anything like this in the airplanes you mention and the simple explanation would be there.  Real Air airplanes do exhibit realistic instability when slow...they can even spin...but they're normally just fine when at speed.  P-factor is also going to be much, much more pronounced in a climb and during slow flight...not at cruise.

[JYW 2,965]

/\ /\  Gregg has a very good point there.  It could actually be noise or lack of calibration of your hardware that's interfering.  One of the weaker points of FSX and P3D is that yaw and co-ordination in general isn't modelled to well, and you often can't tell whether you're in a plane with a 2 axis AP or 3 axis....  so for you to be seeing such severe yawing issues does suggest the problem may be other than the APs, and hardware calibration is a good first place to look.

[aurel42 10]

Hey Gregg, I appreciate your response, but, no, calibration is not an issue. I use FSUIPC for all axes, which makes it easy to verify that there are no spurious inputs, and I verified that the rudder actually rests nicely in its small dead zone when it's not used.

 

Thanks for pointing out that P-factor is much more noticeable when the aircraft is slow, I was aware of that, but it didn't "click" until you mentioned it: When I took off last night in a Twin Otter, I got almost carried off the runway and needed rudder at full left deflection to get myself back to the center line. I attributed it to a strong gust of wind, but looking back now, it could well have been another piece of evidence pointing to a ridiculously strong P-factor setting.

 

I'll verify that by testing in calm wind conditions.

 

I need to find out whether there's a way to get the current numerical value for the P-factor setting (instead of just looking at the slider in the Realism Settings dialog), perhaps there's an FSUIPC offset or something. Or maybe the changed value is visible in prepar3d.cfg.

[Gregg_Seipp 565]

 

 

Thanks for pointing out that P-factor is much more noticeable when the aircraft is slow, I was aware of that, but it didn't "click" until you mentioned it: When I took off last night in a Twin Otter, I got almost carried off the runway and needed rudder at full left deflection to get myself back to the center line. I attributed it to a strong gust of wind, but looking back now, it could well have been another piece of evidence pointing to a ridiculously strong P-factor setting.

 

Left rudder input in a prop plane is a hallmark of P-Factor (and torque).  Did you try turning it to its lowest setting in Settings for a test?

[Bobsk8 12,489]

Left rudder input in a prop plane is a hallmark of P-Factor (and torque).  Did you try turning it to its lowest setting in Settings for a test?

 

 

All  planes I have flown with clockwise rotating  props require right rudder not left rudder. 

[aurel42 10]

Left rudder input in a prop plane is a hallmark of P-Factor (and torque).  Did you try turning it to its lowest setting in Settings for a test?

 

I've set my Realism settings according to the user guides for the RealAir aircraft, which recommends setting the P-factor and Torque sliders to about 50%. My current working theory is that "something" (possibly an addon like FsPassengers) modifies these values without my intervention, perhaps increasing one or both to more than what a slider at 100% would represent.

 

 

All  planes I have flown with clockwise rotating  props require right rudder not left rudder. 

 

By now, I'm quite positive that, when I have the problem we're talking about, I need more left rudder than usual to stay on the center line during take-off, and I need more right rudder than usual to keep the aircraft from going into a slip (and swinging back to the left) during right-hand turns (no matter whether it's an AP-initiated turn or whether I'm in control of the aircraft).

 

If this is the opposite of what's to be expected with "normal" values for P-factor and/or Torque, perhaps that could imply that one of the settings is – by mistake – set to a negative value?

 

I'll try to check tonight whether the behaviour is actually caused by one of my addons (by disabling them all). If I can't figure it out, I'll make a short video demonstrating the issue, perhaps it's easier to identify what's going wrong if you can actually see it yourself.

 

Thanks to you all, I really, really appreciate your help!

[Gregg_Seipp 565]

 

 

All planes I have flown with clockwise rotating props require right rudder not left rudder.

 

Me thinks I can't tell which foot is which...LOL.  Yep, you're right.  I don't even think about it much anymore.

[aurel42 10]

First, I have to correct wrong and confusing info I have given before, sorry for that:

• being carried to the right during a take-off with the Twin Otter was unrelated to the issue discussed here; it probably was just a strong wind, as I assumed initially;
• P-factor and torque don't appear to be part of the problem, especially during take-off and slow flight, the behaviour is as expected (slight left-turning tendency);
• add-ons, especially FsPassengers, don't seem to have anything to do with the issue: I could reproduce the problem after disabling all non-essential addons (using SIMstarter NG, I disabled everything except gauges, FSUIPC and the GoFlight input bridge in exe.xml and dll.xml)
• the behaviour is not asymmetrical, that was a total brain fart; the behaviour is identical during left and right hand turns;
• the behaviour is identical when hand-flying the turns;

Here's the video showing the issue, please ignore the audio track (I like to listen to podcasts while flying and didn't turn it off when I recorded the video):

 

https://youtu.be/NJbwKYGQL9A?t=1m27s

 

• at 1m27s: AP in HDG and ALT hold modes; close-up of instruments during 90° right-hand turn; 
• at 1m40s: ball in turn coordinator swings right, then left
• at 1m50s: ball in turn coordinator swings right, then left
• at 2m20s: similar turn, now with YD "cheat", aircraft behaves as expected
• at 3m20s: one more turn without YD, now in AP's NAV mode, aircraft behaviour back to weirdness

If you have any idea what could be going on here, please help!

 

Cheers,

Marc

[JYW 2,965]

Hi - That behaviour looks quite normal for AP turns without yaw damper.   It's the Yaw Damper that would co-ordinate those turns.

[aurel42 10]

That's totally not what I expected you to say.

 

I guess I'd have to reinstall P3D to show that this aircraft doesn't behave that way on a "fresh" sim.

I've been there before and what I consider an "issue" was gone after the reinstallation until it reappeared recently.

[Gregg_Seipp 565]

Do you have a weather program running?  I'd expect that kind of behavior could happen in the wind.  The wind kicks the airplane and the autopilot lags to catch up.  On the other hand, it looks like the autopilot forgets what it's doing and then wakes up and over-corrects. 

[aurel42 10]

Do you have a weather program running?  I'd expect that kind of behavior could happen in the wind.  The wind kicks the airplane and the autopilot lags to catch up.  On the other hand, it looks like the autopilot forgets what it's doing and then wakes up and over-corrects. 

 

Yes, I'm running AS16.

 

There's a small, yellow pointer at the top of the map on the left (provided by Little Navmap) that shows the current wind conditions. While recording the video, the wind was at a mere 3 knots.