Autopilot Tweaks - NAV and Glideslope

[JerryH 1]

To all FS2004 autopilot users:Ever since starting with FS2002, I've been frustrated by the way most autopilots capture a VOR radial, then overshoot from left to right for miles and miles. My alternative is to use heading hold and adjust as needed to stay on course. However, when the same overshoot problem occurs while capturing an ILS localizer, I didn't think there was any way to fix it.Then a thread of messages started here in late February and ended on March 4th. The subject, "FS9 Approach Capture Tweak", discussed some of these issues but never came to a conclusion that was shared with the forum. But something was suggested that I had never been aware of before. In FS2004 there are 10 control variables in the aircraft.cfg under the autopilot section that the user can adjust. These AP controls (listed below) are described briefly in the FS2004 SDK, but the descriptions are not very helpful. The SDK calls them standard PID feedback controllers (Proportional, Integrator, Derivative). Anyone wanting more details on this subject can look for a Wikipedia entry, "PID controller". That article explains it reasonably well, but it helps to have a technical background. AP Controls Common to Most Aircraft:nav_proportional_control= 9.00nav_integrator_control= 0.25nav_derivative_control= 0.00nav_integrator_boundary= 2.50nav_derivative_boundary= 0.00gs_proportional_control= 9.52gs_integrator_control= 0.26gs_derivative_control= 0.00gs_integrator_boundary= 0.70gs_derivative_boundary= 0.00Most of the add-on aircraft in my hangar came with the above control values, with only a few exceptions. I don't know their origin, but they were apparently just copied from one aircraft.cfg to another with the assumption they must be okay. But they don't work well for most aircraft, particularly when capturing a VOR radial or ILS localizer.The values in the next list are what I recommend today for improving the NAV and glideslope capture of most aircraft. They're the result of a lot of testing on many aircraft. Because it's not possible to have one set of values that are perfect for all situations, these are a compromise. They work well for no-wind approaches as well as crosswinds from either side. At this time, I've tested a total of 23 aircraft and have found this set works well for all but one.Recommended AP Controls for Most Aircraft:nav_proportional_control= 20.0 nav_integrator_control= 0.45 nav_derivative_control= 0.00nav_integrator_boundary= 0.50 nav_derivative_boundary= 0.00gs_proportional_control= 14.0 gs_integrator_control= 0.65 gs_derivative_control= 0.00gs_integrator_boundary= 0.70 gs_derivative_boundary= 0.00From what I've learned, I believe any aircraft that came with the first set of AP controls will benefit from the recommended set of controls. I've tested two other aircraft that came with a different set of "customized" AP controls, the default B737-400 and a Sud Caravelle. The B734 works better with my recommendations, but the Caravelle does not. Because of these mixed results, I can't draw any conclusions about other control sets, including most payware. For payware owners, please feel free to do your own testing and report your results on this thread. There is one open question that I haven't been able to resolve throughout this study, and I'm hoping someone can enlighten me. Many, but not all, aircraft do a small pitch-up maneuver at the point where they intercept the glideslope. As I recall, FS2002 had none of this, but it happens with a majority of my FS2004 aircraft. Changing any of the five glideslope controls seems to have only a small effect, if any. This pitch-up makes it all but impossible to avoid overshooting the glideslope.Be sure to backup your aircraft.cfg file before trying out these new control values. My recommendations don't come with a guarantee, but I think most aircraft autopilots will be improved. Regards,JerryH

[Guest belga1]

Hello,Thank's for posting.I make a test with the tweaked autopilot:LocationMauritius International AirportPlane usedProject Airbus A320-214No windNote:It's a plane and airport I use many time.Result:At least with this plane I don't report any improvement or any degradation in the following of the ILS .. compared at was I'm used to see .. using the autopilot not tweaked.I made a quick small video (not allways informative as you like .. but ... )Small VideoRegards.

[JerryH 1]

Thanks belga1, I appreciate your feedback.I've read good things about Project Airbus, so maybe they had already worked the autopilot problem.I would be interested in seeing the original autopilot numbers for comparison with mine.Regards,JerryH

[Guest belga1]

Hello,There they are:[autopilot]autopilot_available=1flight_director_available=1 default_vertical_speed=1000autothrottle_available=1 use_no_default_pitch= 1use_no_default_bank= 1pitch_takeoff_ga=18 autothrottle_arming_required= 1autothrottle_takeoff_ga= 1autothrottle_max_rpm= 96.5default_pitch_mode=0default_bank_mode=0max_pitch= 10.00max_pitch_acceleration= 2.00max_pitch_velocity_lo_alt= 4.00max_pitch_velocity_hi_alt= 1.50 max_pitch_velocity_lo_alt_breakpoint= 20000max_pitch_velocity_hi_alt_breakpoint= 24000max_bank= 25.00max_bank_acceleration= 2.0max_bank_velocity= 3.0max_throttle_rate= 0.30nav_proportional_control= 12.00nav_integrator_control= 0.05nav_derivative_control= 0.01nav_integrator_boundary= 0.5nav_derivative_boundary= 0.35gs_proportional_control= 15.0 gs_integrator_control= 0.30gs_derivative_control= 0.00gs_integrator_boundary= 0.7gs_derivative_boundary= 0.00yaw_damper_gain= 1.00So.. using those numbers or yours .. I not see any difference in the plane behaviour when intercepting ILS or in the approach (it was like same trim and heading corrections)Regards.

[JerryH 1]

Hi belga1,It looks to me like someone at Project Airbus already did the job of tweaking those numbers long before I started hacking away.As I said in my original post, if the aircraft came with the first list of controls, then my recommendations should improve the autopilot. I learned that it's not possible to improve every AP with just one set of tweaks.It's interesting to note that your glideslope numbers are similar, but the Nav numbers are not. Yet, the aircraft performs about the same. Go figure.Thanks again,JerryH

[michal 44]

I don't think there is any significant improvement when you take a typical FSX autopilot. They all perform poorly in horizontal and vertical axis. Only some aircraft developers (like PMDG) develop their own fully customized autopilot totally independent of MSFS and hence performing significantly better.

[JerryH 1]

Thanks for your comments, Michael. I don't doubt that many of the payware producers provide an excellent autopilot with their aircraft. My post is mostly intended for the excellent freeware aircraft that needed some AP tweaking.Example: One of my favorite 4-engined propliners came with the "standard set" of AP controls from my first list. While capturing the KSEA 340 radial at a DME of about 100 nmiles, it overshot the radial and ended up at a heading (010) that was 30 degrees off course. Then, while trying to correct, it overshot in the other direction until it was 18 degrees off course.Repeating the exact same flight with my recommended numbers, the propliner overshot by only 4 degrees, then smoothly captured the radial with only a slight deviation. I would like to think that was a significant improvement. My example above is not intended to be critical of any freeware providers. They've been providing me with superb add-ons for many years. My only intent is to give something back for a change. Regards,JerryH

[JerryH 1]

Back in early March, I know this subject was of interest to Opa. So I'm bumping it up in case he and others missed it.Regards,JerryH