I've been searching for updates/performance improvements to the default FSX autopilot but have not been able to locate anything. When I look at the autopilot section of aircraft.cfg files, it seems the basic control parameters that determine the autopilot flight performance are always essentially the same. It even seems the autopilots in the payware a/c I have are built using the same default FSX autopilot parameters in their aircraft.cfg files.   As a result, despite looking like and simulating the operating modes of real world autopilot systems to some extent, the actual autopilot flight performance, such as the degree of "hunting" when intercepting and capturing a localizer, etc., is about the same as the basic default FSX autopilot, at least in my limited experience. 

 

If there are any default autopilot flight improvement updates out there, I'd appreciated hearing about them.

 

Thx,

Al

The default AP values for the PID controller work slowly, but they workfor just about any airplane. If you want to improve the altitude and heading capture of the autopilot, you'll have to do it individually for each airplane as the AP is just as much a slave to the airplane's flight behaviour as you are.

I understand your point, but it seems to me that is exactly the reason we have aircraft.cfg files -- so we can adjust/tune values for each particular type, or at least class, of a/c.  It would seem this would be particularly true for payware a/c. Perhaps some tuning of the AP happens other than via the parameters in the aircraft.cfg file, but it is not evident to me looking at the performance of the a/c I happen to have.

 

Thanks for the response,

 

Al

The aircraft.cfg values are all there is for configuring the autopilot.

 

Here's the documentation for them in case you don't know it already:

https://msdn.microsoft.com/en-us/library/cc526949.aspx#mozTocId942782

Yes, thanks, I'm aware of the MS aircraft.cfg doc. One thing that's interesting is in every FSX aircraft.cfg file I've looked at the AP derivative parameters are always zero. Makes me wonder if that aspect of the PID was actually implemented. You would think information related to the rate of change of the position error would be useful information, but apparently the FSX AP does not make use of that info. I'd sure like to see the  algorithm the FSX AP PID controller implements.

Al

As the documentation states, the autopilot controller is of the PID type, so it can be configured in any way that is permitted by a real PID controller.

My theory is that the Aces team simply found a PI logic to work best for the autopilot during playtesting and thus set the derivative coefficient to zero. Hence, there's nothing stopping you from trying out an actual PID configuration in an aircraft.

Yep, I've been experimenting with various PID parameter values. So far I haven't found any values for the derivative parameters that seem to make any difference at all, which again makes me wonder if the "D" in PID was implemented. It could be I just haven't hit on "sensitive" D values. On the other hand, I have been able to reduce the LOC intercept overshoot by reducing the two nav integrator values. I haven't seen any unintended consequences of doing that -- yet.   But there could be long term steady state error effects, like when tracking a VOR radial for long distances, etc.  A larger integrative term in a simple PID tends to promote a quicker "correction" at the cost of increased overshoot (a larger derivative term tends to counter those effects).

I've been using the default Learjet 45 as my test a/c.

Al