Re: [Flightgear-devel] Autopilot broken, Eric maybe?

[James T. <zak...@ma...>]

On 23 Dec 2009, at 11:42, Heiko Schulz wrote:

> 737-300 (Using the autopilot.panel via F11)- in 19.1 the 737-300 was the airliner with the best autopilot behavior recommended for ILS-approaches done by ap. with my last built from 11/27/2009 the aircraft didn't responded anymore on NAV1-Hold and GS.

Okay, that's a useful data point, I'll take a look at that in the next few days (in between eating too much...). I assume all interactions with the system are occurring through the radios and autopilot *dialogs* (F12 and F11), not the cockpit panel? These are the kind of things that people don't mention, but make a big difference in testing.

Do other autopilot modes work? Do the VNAV modes work, aside from NAV1-GS hold? 

> P.S.:I'm beware of that the actually 737-300 needs a own written flightdirector like the Citation Bravo and other aircrafts from Syd has to be real - but how to write one?

Take a look at Syd's scripts? The core flightdirector.nas scripts are pretty clear, though as always I wish less copying of files into each aircraft went on. If you want help understanding what the scripts do, and how they are structured, I'm happy to (try to) help. (Actually I think the scripts could be simplified quite a bit, but Syd may disagree - he knows better than me!)

The flight-drector issue is complicated because the generic autopilot doesn't provide anything in this area. I have briefly wondered about writing a C++ generic flight director, but I don't think it's a good idea - people who want an accurate flight-drector for a real aircraft will find it much easier to write a specific one in Nasal (to go with a custom autopilot) than try to work with a generic C++ one, I think.

I'd also be happy to start collecting an 'autopilot internals' wiki page, to collect information about this kind of thing in a proper place, and to go alongside my existing route-manager and GPS/FMS internals documents.

Regards,
James