Jim Wilson <[EMAIL PROTECTED]> said: > Finally I spent a few minutes and fed some numbers into the derivative portion > of the PID algorithm we are using and figured out what it does. It looks like > it could be most useful for things like autothrottle. Any control that tends > to accelerate in effect as time passes. I'm not sure, but it might help as > well if the measurements are asynchronous because it seems to force linearity > even if the error is jittery. > > Unfortunately it doesn't help with things like the HDG controls since for the > most part aircraft turn pretty steadily (at a given roll angle). Actually as > Curt noted with yesterday's cvs update, the derivative makes things worse in > the case of the first stage of a two stage controller, because it does things > like overstate the 1st stage output error as the aircraft rolls out of the > turn (and the rate of error correction decreases). > > What we need is something that looks ahead and makes adjustments based on > future error in order to avoid overruns due to the momentum of the aircraft. > > ***** patch description below ***** > > This patch adds a 10 second lookahead specifically for the nav1-heading-error. > I'm wondering if we shouldn't add a class for this specific type of thing to > xmlauto so that they can be user configurable. Or maybe there is a way to > incorporate this ability into the PID algorithm. > > In the meanwhile this makes it possible to get the 747 AP working: > > http://www.spiderbark.com/fgfs/xmlauto-jw-2.patch > http://www.spiderbark.com/fgfs/xmlauto-jw.cxx>
Oops that should have been: http://www.spiderbark.com/fgfs/xmlauto.cxx I made a copy so the bad link will work too. Best, Jim _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
