Autopilot Tuning questions

[lionheart]

Hey all,

I am trying to set my AP to be intune with my present projects performance. It is proving to be a bit ellusive.


On Auto Throttle, the speed is always 'over' the set limit. My ASI is reading 'actual air speed' instead of ground speed. Could this be the cause? I have the max_throttle_rate= at default rate; which is 0.10.


When on the ground, doing a take-off with Auto-Throttle set for say 120 knots, the plane will only do say 80 to 86 knots on the ground, and once airborn, hunts for its set speed of 120 knots. What setting causes this oddity where it is slow on the ground, but proper once off the ground? Can this be turned off?


What setting causes your plane to be 'more reactive' to level off? Right now, she levels off 100 feet 'past' her set altitude, then lazily goes back down to the altitude. Drives me nuts. I have experimented with other autopilot settings from my other planes and none seem to be able to cure this. I was wondering which one 'exactly' causes this to be 'more reactive'.

I have been through the FSX SDK auto pilot section and cannot figure out which of these is the one as the labeling of the settings is slightly vague.


Many thanks for any help or idea's.



Bill

 

[Roy Holmes]

Bill,
I agree the autopilot is badly documented, but I can answer some of your questions.

Autopilot speed hold is either Indicated Airspeed or Indicated Mach. Groundspeed is True airspeed as affected by wind velocity and direction, so it has no part to play in the autopilot. I guess you have IAS and not TAS in the FS Set-up pages.

There is an "autothrottle_max _rpm" setting which typically is 90% so if you take off with that setting you will be slow on the ground. Make it 100% and the take-off performance should improve.

The autopilot adjusts pitch by using trim. Altitude hold mode is achieved by pitch changes achieved through trim changes until it settles down. For example if you have altitude hold engaged and close the throttle, you can (if you have a gauge that gives ELEVATOR TRIM POSITION,degrees) watch the trim go increasingly nose up until you reach the "elevator_trim_limit" in the config file. The strength of trim effect per angular deflection is controlled in the .air file, Cm delta trim. So the ability to hold altitude is a combination of the trim range and its strength. There are PID controller values for glide slope capture in the autopilot section and the SDK relates them to angular displacement from glideslope, but I have to say, since they control pitch adjustments through the use of trim, that they may have some effect on altitude hold capture as well.

I have yet to see a good explanation of how the FS autopilot works, but its nature is such that it must use PID controllers. AI have these for Airspeed and heading control, I assume non-AI have them for more than just ILS control.

This may have raised more questions than it answered which is my idea of a good forum post and you are the ace at that!

Roy

 

[sergio]

The AP is controlled by the PID in the aircraft.cfg, section [autopilot]. In my database I found a posting by user "Arthur" in the old (now no more available) AV History forum. Perhaps it may help.

Integrator and derivative control, amongst others, are related to control theory. Integrator and derivative are gains (depending on how you look at it). For example, how well a motor places a needle in a hole is dependent on how well the motor is controlled (and how it is hooked up to the needle). Usually a motion controller controls, or sends output signals to an amplifier. Then the amplifier sends either a voltage or a current to the motor depending on the type of control being used. The output from the amplifier maintains a relationship to the input from the motion controller.

Proportional (how fast and stiff the motor moves the needle), derivate (will cause the motor to oscillate before it gets to the hole), integral (effects error signal which is the difference between the command signal and the actual path taken), velocity feedforward (positive gain that works with derivative), and friction feedforward (another positive gain that helps if a system has a lot of friction in it) are all examples of gains. Adjusting one of these effects the way output siganl going to the amplifier. To adjust these is pretty straight forward once you understand what each one does and how they are related. Usually when you tune motors, or adjust the above mentioned gains to maximize the performance of the motor, you will work with a graphical representation of the output to the motor and then some variable from the motor, ususally encoder counts or error signal, to track the motors performance. If the motor is tuned pretty nicely you will see the motor track the output signal of the motion controller.

So to answer your question, I haven't played with the NAV integrator/derivative or similar things yet. I am waiting for the SDK to find out what exactly is going on. That is assuming the SDK will have the correct information in it.
-Arthur

Ciao, Sergio

 

[lionheart]

Thanks Roy for the input.

There is an "autothrottle_max _rpm" setting which typically is 90% so if you take off with that setting you will be slow on the ground. Make it 100% and the take-off performance should improve.

Roy

Presently the bird is at 150%, so it must be another setting, or perhaps hard wired. I'll keep looking.

The autopilot adjusts pitch by using trim. Altitude hold mode is achieved by pitch changes achieved through trim changes until it settles down. For example if you have altitude hold engaged and close the throttle, you can (if you have a gauge that gives ELEVATOR TRIM POSITION,degrees) watch the trim go increasingly nose up until you reach the "elevator_trim_limit" in the config file. The strength of trim effect per angular deflection is controlled in the .air file, Cm delta trim. So the ability to hold altitude is a combination of the trim range and its strength. There are PID controller values for glide slope capture in the autopilot section and the SDK relates them to angular displacement from glideslope, but I have to say, since they control pitch adjustments through the use of trim, that they may have some effect on altitude hold capture as well.

Roy

That might be it. I'll play around with trim effectiveness and see if that will make it more snappier.

This may have raised more questions than it answered which is my idea of a good forum post and you are the ace at that!

Roy

lolol... yes... arrgh.

 

[lionheart]

The AP is controlled by the PID in the aircraft.cfg, section [autopilot]. In my database I found a posting by user "Arthur" in the old (now no more available) AV History forum. Perhaps it may help.



Ciao, Sergio

Many thanks Sergio. That is quite a good find you have. I am copying it to my gauges notes.

Years ago, in my younger days, I worked at a machine shop, running CNC equipment. Tons of controlled motors that responded well to the auto-programs. I could invision what this man Arthur was talking about. He explains it well. A very good find...


Bill

 

[mgh]

There are two sets of PIDs in the [autopilot] section. The first set is for lateral navigation and is used for reducing the deviation from a course/heading (nav). The second set is for vertical navigation and is used for reducing the deviation from the glideslope (gs).

As an experiment, I changed the value of gs_proportional_control for the default FSX C172 from its default value of 9.82 to 100 (a factor of 10) which made no difference. I'd set the autopilot altitude to 1000' and, in both cases, the vertical speed began to decrease as the aircraft passed 880' and the overshoot was to 1020'. That suggests that it's only relevant when tracking the glideslope.

There is a variable max_throttle_rate which sets the maximum rate at which the autothrottle can change the throttle setting.