[BUG] Rotation/lift off is too quick

[sef3613]

Mod Version

Latest custom FBW dev verison.

Describe the bug

While the flight model is much better with the custom FBW, the plane "jumps" off the ground very quickly right as rotation starts. Additionally, the rotation of the nose seems too quick with too little sidestick input required.

To Reproduce

1. Do a normal takeoff in the custom FBW version.

Expected behavior

Nose slowly raises and eventually, airplane lifts off.

Actual behavior

Nose quickly raises and airplane lifts off almost immediately.

References

Comparison with FSLabs. Same conditions, weight, and CG. Note that the FSLabs requires sidestick input up to about 10 degrees to get the nose off the ground whereas the FBW starts rotating with about 5 degrees of sidestick deflection. Also note that the FBW is at 60 feet RA just when the FSLabs achieves a positive rate of climb.

https://www.youtube.com/watch?v=BWKbU_9yIuE&feature=youtu.be

Additional context

Was this working before/when did the issue start occurring? Always an issue. Is this a problem in the vanilla unmodded game? Yes

Discord username (if different from GitHub):

[hotshotp]

hello - this has been a reported issue with legacy flight model. when changing to modern flight model the issue is mitigated. are you on modern or legacy?

[sef3613]

@hotshotp Modern flight model was on in that video. It's better than the rotation without custom fbw but still too quick and "light" as you see

[derl30n]

duplicate of #2587

[hotshotp]

duplicate of #2587

But that issue was wrong though - we confirmed with real pilots

[MapleNAV-TCASFAULT]

The elevator authority right now without dampening is too aggressive I find, even with the custom FBW, not being cautious can easily stall or severely oscillate the aircraft on takeoff (and yes I’m using the modern flight model), so I do agree with @sef3613, the rotation characteristics of the a32nx is a bit scuffed.

[sef3613]

Its not just the rate of rotation, but also that the main wheels lift off the ground almost instantly, like a C172 takeoff. If you watch the FSlabs or any real airbus video, you can see that the main wheels don't come off the ground until a fairly significant nose up pitch.

[donstim]

One needs to be very careful comparing two different airplane/engine combinations being flown on 2 different simulator platforms. There are differences in controller implementations (sensitivities, etc.) as well as how ground effect is modeled. I understand and appreciate the comments, but we choose to work off of substantiated reference data.

I believe there are two aspects at play here: (1) the airplane pitch rate, and (2) the rate-of-climb away from the runway. Although the initial pitch rate can be very dependent on the pitch input and the controller responsiveness in each sime, one can use the the video to examine the pitch rate. Both airplanes began their rotation at about 43 seconds in (maybe a fraction of a second later for the FSLlabs one). The A32NX reaches 10 degrees at 47 seconds, while the FSLabs gets there at 48 seconds. 10 degrees in 4 seconds is a 2.5 deg/sec pitch rate, while 10 degrees at 5 seconds is a 2 deg/sec pitch rate. Airbus recommended procedures call for targeting a 3 deg/sec initial pitch rate.

The first second after liftoff, the pitch rate for the A32NX looks like it is a bit high, while the FSLabs looks like it might be a bit on the low side. But, at least as flown in the video, the A32NX pitch rate dampens quickly within the next second or two, although again, all of this is very subject to pilot technique and controller characteristics within the respective simulator.

For the very initial part of the climb away from the runway, we are somewhat at the mercy of the ground effect hard coded into MSFS. W don't know yet exactly what is in there and how best to accommodate it. What I will say is that the takeoff speeds, stall lift margins at those speeds, and climb rates are well matched to the real airplane performance data.

In summary, in order to make any changes to the current mod in this area, I think we would need to have substantiated reference data with specific performance targets.

[sef3613]

@donstim I understand, and I agree that the MSFS ground effect is likely what is causing the "C172-style" lift off on the mod. The issue I have is the initial rotation while still in ground effect. I know the FSlabs is in a different sim, but I've heard countless pilots praise the FSLabs flight dynamics and they say it rotates, flies, and lands very close to the real airbus. I know those aren't "numbers" but they are the opinions of many people who have thousands of hours in the real plane. I just find that the FBW feels like it weighs <5,000 lbs when in ground effect and then feels like a proper weight once it is out of ground effect. If this is something only Asobo has control over, I hope they listen to the FBW team and keep working on it. Love the mod, great work!

[sef3613]

Sure, the FSLabs was under-rotated but do you really think the right video (MSFS) looks like an airbus rotation? It springs off the ground instantaneously. I know the FSLabs isn't perfect according to the manual, but just logically watch the video and think "does that really look correct?" for the one on the right. And I'm only talking about the initial rotation until about 20 feet RA

[sef3613]

And the little nose drop in MSFS was the nose dropping as I left ground effect even as I was holding back pressure.I think ground effect in MSFS is a little over-exaggerated still.

[MapleNAV-TCASFAULT]

I was watching V1’s stream a few days back and he was caught of guard by how sensitive the aircraft was on rotation. https://youtu.be/RYnAe1vTM3o @39:31

[sef3613]

Yes, you CAN gently rotate the FBW mod but you have to use far too little side stick input if you have your sensitivity set to feel good during normal flight in the A32nx. So if you use muscle memory from the real plane or even another sim, it seems that it's a common theme to have a super fast and springy rotation.

[sef3613]

I know that's not data from a manual but I think if many people are finding it is too quick compared to what they're used to in other sims and rw, maybe it's worth looking into.

[lukecologne]

Keep in mind that this can be caused by the FBW. On ground, the sidestick controls the elevators directly, but on takeoff above 75kts the maximum elevator deflection is reduced from 30° do 20°. It could be that the default MSFS FBW doesn't do this, so we should revisit this once we have our own FBW system.

[derl30n]

I was watching V1’s stream a few days back and he was caught of guard by how sensitive the aircraft was on rotation. https://youtu.be/RYnAe1vTM3o @39:31

Don't wonna say anything offensive but looked like he used improper technique, pulling on the stick, letting stick completely go, pulling again and centering again.

Just to give an example, The FSLabs320 takes off differently than the FF320.

FF320 heels and behaves similar to our 320. And I would generally go with X-Plane flight model over anything else.

[derl30n]

Keep in mind that this can be caused by the FBW. On ground, the sidestick controls the elevators directly, but on takeoff above 75kts the maximum elevator deflection is reduced from 30° do 20°. It could be that the default MSFS FBW doesn't do this, so we should revisit this once we have our own FBW system.

agree. However we might need a little bit of tweaking of possible on our own FBW since this is also reported on our FBW. But custom FBW makes a very good job and basically removes a ton issue that the default FBW has.

[sef3613]

@MisterChocker yeah I've heard multiple airbus pilots say that the FSLabs has a more realistic flight model than the FFa320 and toliss.

[derl30n]

You also need to keep in mind the real side stick simulates forces. I doubt any here has such a system at home. So the feel is completely off. There is no comparison of the feel from any PC sim side/joy stick airbus to the real deal.

[sef3613]

@MisterChocker No, but you can still set sensitivities to feel correct. I'm telling you, the FFA320 should not be a comparison for rotation as it is really off too

[MapleNAV-TCASFAULT]

Wait what, the a320 sidesticks have force-feedback? I thought only the a350 had such feature.

[sef3613]

Also apparently the Warthog feels very similar in force to the real side stick which is what V1 Sim uses

[derl30n]

We haven't taken any other sim plane as comparison we took real life data to develop the flightmodel.

[sef3613]

Well, the real life data has led the a32nx to have a a rotation that's unrealistic

[derl30n]

Dude are you listening to us?? We simply cannot model the rotation, the model doesn't care about the sidestick, it uses AOA nad generates different lifts at these...

[sef3613]

Also the a320 has no force feedback lol

[MapleNAV-TCASFAULT]

Well I did make a PR before that got reverted to fix such issue, I can possibly open it and refine it more following true airbus performance charts.

[sef3613]

It feels almost identical to the Thrustmaster Warthog hence why so many irl airbus pilots use the warthog. No force feedback

[derl30n]

This is getting way too heated here..

[MapleNAV-TCASFAULT]

AFAIK, the sensitive pitch is due to the pitch authority, as when I turn off all f-ctrl computers on the left side (which should give me direct control) the aircraft is way too responsive to pitch inputs, and it feels even like a Cessna sometimes.

[sef3613]

I mean, I think a large part of the issue is the MSFS ground effect as I was told above. I hope they work with you guys and/or other devs to fix it

[derl30n]

Okay I am playing this card and bringing in the expert. I am summarizing the issue for you so you don't have to read any comment in here since they are very repetitive, but feel free to do so. The issue seems that the response of aircraft in relation to sidestick deflection is unrealistic and wrong. More precise the pitch rotation. I know we've had this discussion internally where we came to the conclusion that this is very dependent on the hardware and software settings for the axis and we may need a config to solve this.

@aguther can you please give as an answer to the open questions and assumptions made in this issue?

I know you have very little time for things like this, I would highly appreciate the effort.

I am most likely going to lock the conversation since the discussion is too heated, please feel free to go to unlock if it's inappropriate. Discussion like that should take place on our discord sever and not on github.

[tareksabet]

@sef3613 so what's appearing from your issue that the plane lifts off instantly on rotate speed.

But this issue has lots of factors such as the CG of the aircraft, weight of the aircraft, your V speeds, your trim and your controller settings.

So to check this issue I will do an extensive testing on it with another pilot with all possible configurations and check it.

But what I saw in your comments above that you compared this plane which is an A320Neo to the FSLabs A320 which an A320Ceo so lots of deferences.

To finalise, I have a question for you which is : what was the CG of your plane, its weight, your V speeds and the trim you set ? Answering this question would help us when we will test it.

Thanks for reporting this issue.

Tarek

[sef3613]

I don't remember exactly, but the cg and weight of both planes was identical. I believe the CG was 29.1 with the trim set accordingly, and the gross weight was roughly 147500 lbs. I used the v speeds from the FBW MCDU and the FSLabs aoc performance which yielded almost identical numbers. Again, I'm not trying to say the FSLabs is perfect and I understand that it's a CEO, but thank you for noticing that the FBW also doesn't seem completely correct.

[sef3613]

Also, my apologies for sounding heated @MisterChocker . I didn't mean to get heated but I realize my wording and tone came across that way over text. I believe that the FBW flight model is very good so perhaps the issue I'm experiencing is an issue with the MSFS ground effect and not the airplane's flight model. Like I said, I really do love this mod and I think you guys are the single best thing happening in MSFS right now. Cheers

[donstim]

I've done some looking into what parameters that we have control over (not considering our own FBW) that affect the lift-off after rotation issue: including varying the elevator size, horizontal tail size, elevator effectiveness, the pitch stability, lift coefficient, and controller sensitivity settings. The tests were subjective with no recording of any data, so take accordingly. Also, I only conducted a couple of takeoffs with each setting that I tried, and these were at the default weight and cg.

One of the more interesting findings was that changing the horizontal tail area and elevator area had no effect at all. Even with values of 0, there was no discernible change in takeoff flight characteristics.

I felt that changing the lift coefficient was somewhat effective, but it affects the behavior more after leaving ground effect than while in ground effect. With the current state of what I know about the flight model, it would be difficult to make a change to improve the takeoff flight characteristics without adversely affecting other areas that are now well tuned. Part of this is due to the incorrect treatment of slats by the MSFS flight model. However, if warranted, it is something I could continue to pursue.

Changing the pitch scalar had little to no effect. Changing the elevator effectiveness can affect this issue, but there is so much more elevator deflection available than is actually needed, you end up just requesting more elevator. Then, you can easily overshoot your pitch target and have a difficult time getting back to where you want to be. The real airplane actually has more elevator deflection available than what is currently represented in the flight model cfg file.

What I considered to be the most effective way to address this issue at the present time is to adjust controller sensitivity settings. I have a Thrustmaster T.Flight HOTAS X joystick for which I had been using the default sensitivity of -50% and no dead zone. I found that setting sensitivity to -70% and dead zone to 7% worked well for this.

[MapleNAV-TCASFAULT]

I also use the extremity dead-zone and reactivity feature that was recently added, I set the extremity dead-zone to 50% and the reactivity to 10% and it was way easier to fly. Note that these settings are terrible with custom FBW (takes forever to flare past the horizon), but they work really well with the default FBW.

[MapleNAV-TCASFAULT]

Ok, sorry to bother with this issue again. I'm right now tinkering with the flight_model.cfg and I have come to an answer, where the aircraft will rotate around 3 degrees/sec with 1/3 upstick input. I reduced the scalar of elevator_maxangle_scalar by 1/2 and the elevator effectiveness to 0.2. It does work pretty well coming in for a stabilized approach, but if you approach too fast, the plane tends to spring back up into the air due to the overdone ground effect (i'm not sure if the ground effect values can be changed or not). Also the edit to the cfg doesn't work really well with the custom FBW branch.

[derl30n]

elevator effectiveness to 0.2.

Reducing this value will have negative effect on in flight performance.

[donstim]

Ok, sorry to bother with this issue again. I'm right now tinkering with the flight_model.cfg and I have come to an answer, where the aircraft will rotate around 3 degrees/sec with 1/3 upstick input. I reduced the scalar of elevator_maxangle_scalar by 1/2 and the elevator effectiveness to 0.2. It does work pretty well coming in for a stabilized approach, but if you approach too fast, the plane tends to spring back up into the air due to the overdone ground effect (i'm not sure if the ground effect values can be changed or not). Also the edit to the cfg doesn't work really well with the custom FBW branch.

Are you saying ground effect is worse with these changed values? Maybe because it forces you to use more elevator deflection due to the reduced effectiveness?

As far as max elevator deflection angle goes, I don't think that will have much effect because hardly any deflection is normally needed (maybe a quarter of what is available) but maybe with the reduced effectiveness it may come into play? Shouldn't for the flare, but maybe for takeoff rotation?

[derl30n]

Flight Model meta issue for tracking #3105

[aguther]

One additional info: I discovered that the neo actually has a rotation law. I put this on my list to be implemented for the custom FBW and it will streamline the rotation performance because it’s basically a pitch rate law.

[donstim]

With the difficulty I am having in trying to find an acceptable aerodynamic solution, the custom FBW sounds like the best bet. It would override any solution I found anyway (but I'm still looking!). I am glad the airplane has FBW and doubly glad that you are implementing an accurate version of it!

[aguther]

A first implementation of the rotation law together with improvements in the C* law are now available on the fbw branch (can be installed using the new installer and choosing the experimental version). Please try it and note any feedback here.

[aguther]

The rotation law got positive feedback so I will go forward and close this issue. Feel free to reopen it when you think the rotation law does not solve the problem.

[lionhell]

"Feel free to reopen it when you think the rotation law does not solve the problem."

Hi ! Since few days/weeks I'm getting problems with rotation and A32NX dev version. Formerly I got no problems, all was fine, never broke airplane at rotation before. But now I'm scared every time I takeoff ! Don't know the reason but let say 1 on10 times, at rotation plane is overstressed and breaks ! Very annoying because I don't know why and more over... it is not everytime and never happened before ! Sor for me situation is worst than previously (let say FS SU5)... any clue ?

[aguther]

No idea, we did not change anything in that.

[aguther]

Can you provide me with an FDR file when this happened?

[donstim]

Sure would like to see a video of this. I've tried numerous times at different weights and speeds and cannot pull nearly enough g at rotation to overstress the airplane.

[lionhell]

OK I’m « lucky »… on the 1st attempt the plane broke at rotation 😊

I sent you the video with « grosfichiers.com » I hope you received it.

Regards,

Lionel

---

https://home.mcafee.com/utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient

Scanned by McAfee https://home.mcafee.com/utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient?utm_medium=email&utm_source=link&utm_campaign=sig-email&utm_content=emailclient and confirmed virus-free.

[donstim]

Where did you send the video? Can you just post a link here?