Closed jan-hudec closed 9 years ago
pedromorgan
commented
9 years ago Can u supply more info..
1) what is the vertical speed 2) what "pitch up" does the aircaft kinda get to when it stalls.. 3) what altitide...
My understaning is that most airliners on takeoff/climb and "srs" (speed reference system), so after takeoff, the nose is raised or lowered to maintain speed..
pedromorgan
commented
9 years ago oops also mean to add.. Whilst on the runway, can you set the vs to say 2000fpm or 3000fpm and then switch..
jan-hudec
commented
9 years ago Vertical speed exceeded 5,500 ft/min, pitch exceeded 30°, not sure by how much as the AI was all blue. The pitch was added progressively. The plane flew for a while at ~160 knots, just above flaps-up stall speed, and the pitch slowly raised while the speed decayed until the stall warning sounded after which the AP still seemed to keep pulling up. I had to turn it off manually.
Altitude was few thousand feet. However the first time it happened I turned AP off, flew manually for a while and later tried to engage it again at ~FL160 and it again started pulling up until stall.
When I push the altitude knob (which turns the altitude restrictions purple as I'd expect), the vertical speed goes blank. I didn't yet see the problem with the altitude knob pulled.
jan-hudec
commented
9 years ago Even when it seems to work reasonably it does not seem to respect the selected vertical speed and the selected vertical speed keeps disappearing (both managed and selected altitude).
jan-hudec
commented
9 years ago Another problem: This time it climbed fine to FL340 and levelled off, but then it the climb rate kept oscillating between 0 and smaller positive values (+700 or so) and the aircraft kept slowly gaining altitude. When it gained ~1200 ft (FL352) a diverging oscillation in pitch started with progressively. I paused it when the V/S oscillations were about -1000 to +1500 ft/min, the aircraft still slightly climbing on average. I would write down some internal properties, but I don't know which ones.
jan-hudec
commented
9 years ago Ok, the oscillation is clearly problem of the elevator PID controller itself. The /servo-control/elevator-vs is true, /flight-management/control/speed-with-pitch is false, so the FCU/AP Elevator Control VS controller should be enabled, /servo-control/target-vs gradually decreased to -50, yet the /velocities/vertical-speed-fps oscillates between about -15 and +20.
Looking at the controller the Td appears absurdly low (elevator directly controls roll acceleration and the aircraft has a lot of inertia (and I am testing this on A321, which has highest), so some prediction is definitely needed for proper damping.
I am also not sure about the limits. If the plane was properly trimmed, the elevator could be kept within tight limits, but it is not as the FBW on master only trims when all /controls/flight inputs are 0 and they are not.
I tried adjusting the values to <Td>1</Td> and slightly reducing sensitivity to <Kp>-0.005</Kp> and that seems to work fine, though I have to do more tests.
However I have a slight problem with this. The Td of 1 basically means “estimate the input 1 second ahead using its current derivative and use that as targets”. Without working FDM it is needed, because the plane exhibits rather pronounced and rapid phugoid oscillations. However when the FBW is finished, there will be another controller in the chain to take care of that so this one shouldn't predict so much again.
And the same goes for the elevator limits. Without FBW the autopilot needs to be allowed to move the elevator in large range, because the plane may be out of trim and need the elevator input just to remain stable. But the missing FBW controller should be taking care of that, so then the values in A/P will be correct again.
Unfortunately making the FBW is hard. I have looked at it some time ago and didn't get far. The problem is that the A/P controller only needs to be tuned approximately, because it does not matter much if the autopilot takes a second or two to stabilize. But the FBW controller needs to react almost immediately and must not cause oscillations to be controllable by hand and that can't be tuned by trial and error and probably requires speed- and weight-dependent coefficients.
jan-hudec
commented
9 years ago Oh, I see it now. After more testing I realized, that the FBW works in A320neo, but not in A319neo and A321neo. Therefore the autopilot, which is designed to control the FBW, also works fine on A320neo, but not A319neo and A321neo.
The first time I had problems it was with A319neo. Then I tried the A320neo a bit and it worked, but afterwards I picked A321neo and it showed the problem again which prompted me to report it.
I'll look into unifying the set files so that all three use the same, correct, flight control system (and generally eliminate any other differences that don't make sense to be there).
FGDATA
commented
9 years ago Sounds like a good plane Jan
On Thu, Aug 27, 2015 at 11:07 AM, Jan Hudec notifications@github.com wrote:
Oh, I see it now. After more testing I realized, that the FBW works in A320neo, but not in A319neo and A321neo. Therefore the autopilot, which is designed to control the FBW, also works fine on A320neo, but not A319neo and A321neo.
I'll look into unifying the set files so that all three use the same, correct, flight control system (and generally eliminate any other differences that don't make sense to be there).
— Reply to this email directly or view it on GitHub https://github.com/FGMEMBERS/A320neo/issues/6#issuecomment-135509414.
jan-hudec
commented
9 years ago @FGDATA, as you probably noticed, I've already done it and pushed it.
However, I have other issues with the FDMs. Many issues. I'll probably post them as a separate issue for discussion.
This is weird problem, because I am not able to reproduce it consistently.
Every time it happened I had:
The nose kept rising until the aircraft stalled.
But other times I tries it kept climbing and accelerating as expected and I am not sure what I may have done differently.