LNAV Limits?

[Guest]

All,Here's an interesting question: is there a limit for the LNAV's ability to hold the flight plan's track through a turn?On the WYLDD1.DRK DP out of KLAS tonight (I love that DP name!) we departed rwy 19L with no problems, VNAV tracked nicely, and LNAV acquired the track easily. First turn looked good (crs 190 to crs 219 at DEALR), but the system had a hard time through the turns at IDALE (219 to 100) and HITME (100 to 069), which I assumed was because the 100 leg was only 15 NM long...however, at the WYLDD turn (right to 121) it seemed to come unglued. The course flown veered back and forth across the intended track at angles greater than 30 degrees. With each correction, the plane would hold its bank angle until across the track before rolling back towards the track. With each pass across the track, the angular error was greater until eventually, I had to go to HDG SEL and bring the plane back on course. (I was glad for the new door locks...my passengers formed a lynch mob and it was only the wild gyrations of the plane that kept them at bay.)Is it safe to say that any turn more than 30 degrees requires the pilot use HDG SEL to smooth the turn?Kurt KalbfleischAmerica West Virtual

[Guest amerton]

No it would not be (safe) :-)It really depends also on speed. Turn rate at 250 will be greater than at 315. We've had no problems with angles up to 90 degrees as long as the following leg is long enough for the aircraft to recover any overshoot it might have done. anthonyAnthony MertonPrecision Manuals Developmenthttp://www.precisionmanuals.com

[Guest]

Anthony,Thanks for the response. OK. Is it dependent on computer speed? Not that I'm running a slow computer, but I was also running AutoMETAR at the time. Makes me wonder if I'll be able to have real weather along with SB and a realistically flyable plane.As mentioned before, I'm running a P4/1.6Ghz with 256Mb mem, GeForce 2 Ti 64Mb and an SB PCI 512.Kurt

[lefteris 70]

Computer speed has nothing to do with LNAV turns...By "speed" Anthony is referring to kIAS (Indicated Air Speed).Cheers,

[Guest]

Lefteris,Thanks! I got that! But since computer speed has been identified as an issue with climb oscillations, I wondered if it also was a factor in LNAV.I don't think it's a question of physics so much as how the turn prediction works with the new A/P. On the old one, you flew to the waypoint and then made a turn towards the new one...this one predicts the turn rate necessary and starts the turn early to stay on the intended track. (At least, that's how I understand what I see...)What happens in an overshoot situation, where the actual turn puts us OUTSIDE the intended path, is that the A/P continues the turn to intercept the track, and holds the bank until the plane crosses the track. It then rolls back towards the track and turns back into it. Sometimes, this process is barely noticable, and others, like last night, I get a series of increasing angular errors across the intended track. What I'm thinking is that the LNAV ought to roll towards the track until an intercept course is established and then roll wings-level until the plane reaches a point where it can turn back onto the track. I'm not seeing that happen in all cases.Kurt

[lefteris 70]

Well,This has not been seen in beta testing - except in very astute turns, with >90 degree turns, etc.As far as the Autopilot goes:We asked TRI repeatedly for assistance, in order to further fine-tune the Autopilot (.FCS) file. We broke our heads on the keyboard sometimes, to get numbers that would perform close to 95% accurately, 95% of the time. This is the absolute BEST we can do with the internal Autopilot (unless someone at TRI is willing to show us more of the internal structures, so we can tune it better).The next step, if deemed feasible (consider it a LOT OF WORK to rewrite the lateral and vertical PID controllers) is to write our own pitch/roll calculators, so that the whole thing would be entirely tunable by ourselves.Here's a screenshot of the tuning graphs that we have built internally at PMDG, to fine-tune the speed/pitch/roll autopilots:

[Guest]

Lefteris,Thank you, thank you. That answers my question! When I saw it, the problem began with a 119 degree turn.I guess the solution for me as a user is to develop a work-around for now.Sorry I haven't said it yet, but you guys have put out a first class simulator here. Truly incredible work, and I'm very grateful.Kurt Kalbfleisch

[Guest]

Make sure you don't have distance compression turned on when approaching a turn, as the autopilot does not appear to change its calculations about turn rate and the aircraft will not stay locked on the track. Oscillations get much worse with compression on, and the higher it is set the worse the oscillations.--Paul Dickson

[Guest]

Distance compression??? Do you bite your thumb at me, sir? :-lolWe don't need no steengkeeng distance compression.