Some technical details on the 777 AP (English, sorry :-)

[Anthony]

Dear all,

As you guys are very technically oriented, I'd like to shed some light on how the 777 AP is layered over the standard Fly! AP. This will also give you an understanding that you might benefit from in flying the aircraft.

LNAV : the 777 LNAV mode is the current Fly! LNAV. It was not possible, without spending a prohibitive amount of time, to re-write LNAV. Check on www.precisionmanuals.com when I started working with PMDG and you do the maths .
So, along with what is working, come also the weaknessess, such as no hdg change anticipation, badly implemented procedure turns for tights angle legs etc. etc. The Hawker having much less inertia, those problems will not be as severe. Most importantly, no way of editing waypoints and adding / deleting etc. Very frustrating indeed...

VNAV : I had time to rewrite completely VNAV (Fly's VNAV is a sort of VNAV PTH, but as I extended the modes, I rewrote everything, and the VNAV is our own).

VNAV SPD works by setting programmatically the VS to, given a constant CLB- or derated- thrust, maintain a target speed.
I've implemented a good damping algorithm, but unfortunately its usage is limited by a Fly! AP problem : when setting the VS, the ***AP will briefly re-initialize ALL its modes***. Under certain circumstances (slow machines, AC banking), this will cause LNAV to disengage and the aircraft to fly where it feels fit to...
I found out that the optimal maximum timespan to set VS was 1 second. If I hit the AP with too close in time VS sets, it will disengage. This is VNAV SPD will not be too good at reaching and maintaining the target speed. It also explains why VNAV should not be engaged in a bank and sometimes loses the LNAV path. You can see the AP re-initializing its lateral modes when I set VS by watching the FD needles : they flicker to the left or right...

VNAV PTH is better because I don't need as often to set VS, and once set to target next waypoint alt, it will sort of stay there. Also, in PTH mode, the AT maintains speed, not pitch.

FL CH is mainly built upon VNAV SPD logic, so it carries the same problems, i.e. risk of losing LNAV path, no retro-action when banking etc.

I tried two things initially :

1) use Fly!'s AP IAS PITCH mode, which does the job required by VNAV SPD. The problem is that the AC oscillates like crazy. IF you've got Wilco's 737 (pre-last patched), try to climb and press the SPD MCP button, you'll see what I mean. So this solution was a show-stopper.

2) control pitch using the elevator trim, much more precise than the VS speed bug... PRoblem there is that when the AP is on, it takes control over trim setting, and programmatically setting it has no effect.... showstopper as well.

So I *had* to go the VS speed bug way, but it is not without problems.

I would have liked the pitch to follow the same logic as the AT damping algorith I implemented, but it was not possible because of this internal AP resetting thing... I did contact TRI about it, but they answered it was by design, that on small aircraft it was not identified as a problem at the time, and they could not fix it / amend it so near to Fly II coming out. To be frank, as you know, they did quite a massive job fixing SDK issue to help us do our thing.

So, the bottomline is compromise : offer functionnality, trying to reduce the problems that are the result of it.

Also, I lost quite some time in locating and identifying problems, at the cost of some incomplete / missing FMC pages : you don't want to know how many very frustrating hours I spent trying to identify this LNAV getting disengaged. The connection with the setting of VS is so un-related it was sort of by sheer luck I eventually found out.

No need to tell, we're looking forward to implement our own AP in FlyII...

I'm working on a few additional fixes and small additions, like your specific non-US default of thrust reduction alt input in the FMC.

Right-O, have a nice time flying our 777 birdie

Cheers,
Anthony

[Simeon S]

Danke alter Eidgenosse.

Ich schätze neben Deiner Arbeit auch, daß Du Dir die Zeit genommen hast, uns einige Details zu zeigen, die das Verhalten des AP erklären.

Daß Du nicht in Deutsch geschrieben hast, sollte für die, die an solchen Details interessiert sind, kein Problem sein. Außerdem kannst Du ja Deutsch lesen. Ich meine, Französisch zu lesen fällt mir auch leichter, als einen Beitrag in dieser Sprache zu schreiben.



Salut vers Geneve!

Simeon