X-Plane 11 Taildtagger & Stinson erratic absurd ground handling

[soaring_penguin 428]

@Griphos: I like that a pilot with a lot of experience chimes his light. But my understanding in the way Austin works is to write him an email directly, the bug report is being used by the Ben and others on the team. 

But the problem is indeed that you do not give other feedback other then "it is not realistic, the underlying physics are wrong, it must be fixed". I think if you want to get this fixed, is to get into direct discussion with Austin, proving him what is wrong (the sim has the mechanics in place to show airflow, forces and so on in real time). The torque bug got fixed this way, after someone proved mathematically that the calculations were twice what they should be.

I still believe that the underlying system (I say system, not the actual calculations) is correct, but needs verifying and adjusting. Nothing you said proved otherwise (again: your only claim is that it is just wrong, you did not show what is wrong, and by that I mean where the calculations are wrong, not the reaction of the sim). 

On the record, I do believe you if you tell the planes does not react as expected. But you'll have to come with a bit more to convince Austin to overhaul the underlying maths.

[jcomm 5,930]

Well, 

me either ( I mean regarding official experience in taildraggers since, most of you know, I am a glider pilot "only" ... ). Janov OTOH is a pilot with airline experience AFAIK, so, just as every pilot I approach asking about how they find this or that modelled in a flightsim, I get a different answer from the typical simmer only experience.

I'll give an example, with the recently released and bought Condor v2. It now includes 3 gliders I've flown IRL... How do I find it modelled since it's probably the most advanced soaring simulator available ?  Well, I find it nice, useful for task planning and training before going flying for real, but that's about all....

The single experience of taking off tugged in a windy day, or landing, and depending on glider model, makes it look so simplified in Condor, but also in X-Plane after all when I take the K-21, that I'm driven towards a much more relaxed evaluation of the quality / accuracy of a PC-based flight sim... Could I say Condor v2 is crap ? Surely not since the main aspects of glider performance and response to control inputs, within and even outside the normal flight envelope are very concisely and believably implemented.

I say IL-2 is the best, but again, this is the simmer speaking, not the RW ww2 pilot expressing his opinion.... How can I talk about accuracy of the FDM when I haven't really had even the chance to seat on such a ww2 fighter ?

Yet, I've asked some rw pilots, Klaus Plaza being one of them who kindly answered lot's of questions regarding the 109 K-4 as modelled in DCS World... Strangely of all of the youtubes I watched, pilot reports I've read, and use of IL-2, DCS and XP11 109s, the one that probably closely matches what should be the experience of taxing and taking off / landing in a 109 is probably the MDLAG 109s... and it's interesting to know that the author has always made sure that he only uses the instruments made available for aircraft design to implement his creations - Plane Maker and Airfoil Maker.  He's also very receptive to suggestions, and if the initial models didn't use that "control phase-out" to try to mimic control forces, he started using that section in plane maker to add some more feel to operations at higher speeds and loads.

Sometime ago Murmur dedicated some time, and careful analysis, to the effects of winds during ground operations, and he actually made some interesting modifications to the default C172 that, without using any sort of Art Stab, made the aircraft a lot more believable and controllable under x-winds during taxi and takeoff / landings... So, I always have, in the end, to give the benefit of doubt to the true capabilities of X-Plane's lFDM. Maybe when really detailed implementations of all components of an  aircraft model are poured into PM and AFM, and some subtleties are used to overcome the core limitations, we can really end up with very believable results. Again I emphasize that that Grumman AA-5 is a great example of a very nicely modelled Light GA available for XP that appears to be extremely realistic in ground handling and inflight too, just as the MDLAG 109, and I also should say that I enjoy that Corsair F4-U :-)

Edited March 11, 2018 by jcomm

[Griphos 138]

@jh71: You do realize that the things you say I’m saying in quotes were not said by me, right?  You made them up. And the things I actually do say, you appear to ignore. Perhaps you don’t understand the specific dynamics I’m talking about?

Again, it is irrelevant whether the underlying code is capable of modeling the “physics” correctly if it is implemented in a way that models the “physics” absolutely incorrectly.  Until they are modeled correctly, then the code's ability to do so is an item of faith on your part.

What proof do you or anyone else have that the underlying code is correct?  Other than that it models flight behavior correctly?  That’s all any of us have as a basis for judgment.

I don’t think anyone outside Laminar has access to the code, so asking for how the “calculations” are wrong isn't likely to get us anywhere.  That’s like my mechanic telling me he won’t fix my broken steering until I tell him how the tie rods are mistranslating the control inputs because the fundamental steering theory of control linkages is “correct” and capable of turning the car correctly.

And, again, many good developers have talked to Austin directly with clear arguments and evidence that the modeling is badly wrong, and Austin does not listen. That’s Austin. That’s how he is. There isn’t enough money in the world to compensate for the pain in the butt it would be to try to develop for a platform whose primary developer is still radically changing it in its 11th version because he gets some wild hair idea about a new model for some flight dynamic he doesn’t actually understand. 

I don’t understand why you are telling me I need to talk to Austin with some kind of proof about the fundamental or practical errors of his code  I’m not arguing or talking to Austin.  He’s not part of this thread  I’m just pointing out to the author of a video  that he doesn’t appear to know what he is talking about when asserting without any evidence other than a very bad takeoff and landing of a plane that isn’t even designed for the current XPlane that nothing is wrong with the physics, which he inserted into a thread where everyone else is talking about the ridiculous behavior of taildragger aircraft in this sim (without in any way impugning the underlying modeling of the code).  The behavior is ridiculous. I don’t have to prove anything to Austin in order to call attention to the fundamental errors of the video.  The modeling of taildragger ground handling is not correct.  And the only proof necessary for that claim is that the behavior is not correct.  That is obvious in the video itself. 

I fly XPlane almost exclusively because I’m a GA pilot and XPlane models GA flying overall, including the larger flight world environment, much better than any other sim. But I’d very much like Austin to stop breaking my payware aircraft for no darned good reason, making them less realistic in the process because he has what he thinks is a good theory about the “physics” of airfoils.  The point of theory is not to be right. It’s to model the world correctly, which is not happening in this sim in some very important ways. And the way you prove that is not by arguing with the theory. It’s by pointing directly to the behavior. 

Edited March 11, 2018 by Griphos

[Griphos 138]

And good evidence of the bias of the author of the video is his dismissal of the tires dynamics as some kind of “screeching” bug. The sim doesn’t just throw in random screeching sounds, such that some kind of code bug makes them too frequent or loud. There is no screeching from tires unless the code model thinks the tires are experiencing heavy side loads. The code model thinks this, particularly with taildraggers, almost all the time, particularly if there is any crosswind component. 

That, by itself, shows something seriously wrong with the code and its modeling. It’s not a bug. What in the world kind of “bug” do you think that is?  Calling it a bug is just a way of dismissing some departure from correct behavior as not being important.

Edited March 11, 2018 by Griphos

[Janov 2,383]

I stand by my point, the underlying forces calculated are present in real life as well - even if the actual force calculated may not be correct in its magnitude for a particular airplane portrayed. This may be partly due to erroneous assumptions or generalizations in Austin´s code, or to failure of an airplane designer when making the .acf file with planemaker - or both.

I can understand your frustration because your model in X-Plane does not behave like your model in the real world. You may be expecting a bit too much from a general flight simulator if you expect to train your muscle-memory with X-Plane. As a matter of fact it is dangerous to try and hone your skills with an unapproved training device like X-Plane, but that goes without saying, I think. X-Plane is approximating reality and out of the box it is just entertainment software. You can´t expect to be able to practice flying your plane with it.

Yes, the wheels in X-Plane slip too much and screech too early, but this is not something that real tires don´t do. They may do it not as readily, but they do.

Taildraggers ground loop too readily in X-Plane, but they do ground loop in real life, too.

The rudder isn´t as effective as I would expect it to be in X-Plane, but real rudders run out of authority, too.

I have never flown any plane in any flightsimulator (including the big level-D sims they stick me in every 3 months) that feels EXACTLY like the real plane, especially when it comes to take-off and landing.

You can keep posting how "Its all wrong, I tell you, and its all Austin´s fault, and the whole thing is a buggy mess and they really don´t know what the crap they are doing!" - it´s the internet, after all. I think it is not going to help your cause of improving X-Plane´s portrayal of taildraggers much, though.

Jan

[MarioDonick 752]

On 7.3.2018 at 3:04 PM, dts1 said:

Why don't you quit flubbing around with this fundamental problem and hire

If you use the word "you" and the suggestion "hire" to address X-Plane developers: The .org forum is NOT an offcial X-Plane forum. X-Plane developers usually are very rarely there.

12 hours ago, Griphos said:

This isn't a classroom exercise or thought experiment.  It's a simulator.

First and foremost X-Plane is an entertainment software -- a video game, regardless what marketing says. Certain combinations of X-Plane (and ESP-based sims like P3D) with hardware have been certified for procedural training, but usually these fall in the simpler categories AATD and BATD (FAA) resp. FNTP and BITD (EASA), i.e flight physics and feel is secondary whereas cockpit/panel layout and instruments is primary. I don't know if X-Plane has also been used to power FTDs and FFS where other aspects of flight come into play. If a certain combination of software and hardware is one of these official categories is a question of certification of the specific setup. So first of all I would really scale down expectations on what something like X-Plane or P3D or AeroflyFS can do for us, esp. when it comes to real basic flying.

This said, of course we nevertheless want to follow the ideal of "super realistic" simulation, and many developers build their marketing around this expectation. And the ground handling in X-Plane, it is indeed ... well, challenging. Developers are in contact with Laminar about this, but even then it is difficult to narrow this down. And if a solution has been found, a future update to X-Plane may break the solution again. For example, more than once our own aircraft models had flight models working great in one version of X-Plane, but were broken in the next (such as 11.05 vs. 11.10.). Users are expecting (and sometimes demanding) fast "fixes" for these changes, and instead of creating new products (necessary for income -- we need money to operate), we have to constantly tune planes that were actually finished (and I mean this besides bug fixing -- this is a different topic; if there are bugs, we try to fix them. But it's exhausting to chase X-Plane's own changes.)

Anyway...

4 hours ago, Griphos said:

I don’t think anyone outside Laminar has access to the code, so asking for how the “calculations” are wrong isn't likely to get us anywhere.

It's not a question of having access to any source code. It's about reporting bugs properly and showing how to reproduce the bug. And this is not because these stubborn developers don't want to listen. It's because they need as much structured info as possible.

In the example of the ground handling, this at least requires

1. a defined test situation (which aircraft? which airport? which weather? which input hardware used? how many flight models per frame? and so on)
2. your expectations (how SHOULD it behave in your opinion under the situation defined in 1, i.e. how should the ground handling be?).
3. your observations (how DOES it behave in contrast to 2, i.e. how is the ground handling instead?).

The hardest parts are 2 and 3:

• Part 2 is hard because we usually "know" how something should be, without being able to externalize this knowledge (i.e. we have difficulties to convert this knowledge into "objective" data). For example, we are currently developing a C42 (that's a small ultra light on which I learn flying since 2017 -- I passed my written theory exam today, by the way :D the exam took 3.5 hours). Our main developer has never been in a C42 (but has a similar license, and also a PPL), so I give him some basic expectations that add to the official published data of the aircraft. So I tell him how the C42 model should behave, based on my feeling in combination with the data. It's not easy.
• Part 3 is hard because of the same reason. We "feel" when something is wrong with a model, but it's hard to "prove" this feeling with hard facts. When our main developer has finished a new alpha version of the C42, I test-fly it. I then tell him what already works nice, and what works bad. He then improves, and I test again, until we have something that matches my own C42 experience as closely as possible. (And then we let beta testers with C42 experience do the same, and there is no doubt that their experience will differ from mine). But to give proper feedback to our main developer, I can not just say "the airplane is too fast during takeoff roll" or "the flaps are too strong". I have to give him numbers, because this is the only thing the computer can process and what he needs to enter into PlaneMaker's forms.

Thankfully, X-Plane allows to output a lot of data regarding flight model/physics and internal states (datarefs). Some can even be shown visually during the simulation and they can also be exported as detailed tables and log files for later evaluation. The torque bug that was fixed with the help of the community was "proven" using these tools, and I'm afraid that the ground roll issues can also only be fixed with such a structured approach: Expect (physics/behavior), Observe (physics/behavior), Export (data on observations), Report (exports).

Edited March 11, 2018 by MarioDonick

[jcomm 5,930]

Just for curiosity sake at least a commercial / professional class of FNPTs is using X-Plane for Airbus type ( and other aircraft ) solutions. You can search at :  http://www.simlocresearch.com/en/products/simulators

 

[Griphos 138]

Some of you keep changing the terms of the discussion. No one is asking for this to be a certified training program in this thread. No one, certainly not me, is expecting to be able to train muscle memory or practice flight skills with it.  It's certainly NOT usable for that!  And no one is asking it to be. It’s an entertainment flight simulator, yes; and as such, we’re just asking for it not to be wildly inaccurate in some aspects of the flight regime as a simulator. 

Some of you are arguing both that it’s perfectly fine as a simulator AND that we can’t expect it to do a very good job as a simulator. I suspect you’re just arguing to argue now. 

[MarioDonick 752]

7 minutes ago, jcomm said:

FNPTs

Yes, as I also mentioned above, in this class both ESP and XP are used, and specific setups have been certified. But my point in reply to Gripho's emphasis on the term "simulation" was that in this class of FSTDs such things like ground handling are less important than panel layout and instruments. (I did this because "simulation" should be taken with a grain of salt when X-Plane, ESP-based sims like P3D/FSX or AeroflyFS are used on a desktop PC/Mac).

[MarioDonick 752]

2 minutes ago, Griphos said:

we’re just asking for it not to be wildly inaccurate in some aspects of the flight regime as a simulator. 

Some of you are arguing both that it’s perfectly fine as a simulator AND that we can’t expect it to do a very good job as a simulator. I suspect you’re just arguing to argue now. 

If I was arguing just for the sake of arguing I would not have taken a lot of time for sharing our own development experiences.

In your posts, you used bold letters for emphasiszing the term "simulation", i.e. to express that you expect X-Plane to be an accurate simulator. My point was that we maybe should treat X-Plane itself more as a game and use the term simulation only when used with certain models (such as the FF A320 or the IXEG 737) and hardware (regardless if certified or not).

(Because when I read your and the original poster's posts, I had the feeling that you were very upset about X-Plane or Laminar, or disappointed. Basically my posts have the intention to say: "relax, guys. It's just a game." But English is not my first language, so I may attribute emotions to your posts that have not been there.)

[jcomm 5,930]

@Griphos,

I am with you regarding the fact that I would really like Austin to address the current state of the art in ground physics, since in at least one of the recent public sessions he attended he recognized it was far from being ok... But then again, when I look at ground handling in MSFS-derived sims, specially when it comes to taildraggers, it is even inferior, unless some really special out of the core techniques are used... 

That's why I mentioned IL2, and in some way DCS, both flight simulation platforms I ended up joining for quite a while although I knew they'd be completely useless for me since it's the type of "flight" i really don't even, by far, like... although I love some of the aircraft... I went into those platforms to find what I can't find in ** any ** of the civil flightsims I had used before...

But all summed up, and depending on the use you're willing to give to the sim, we have to accept it as it is and keep looking fwd for it to be addressed. At least in X-Plane we can have some more hope that it hapens, if not in XP11, maybe by XP12...

I use ELITE IFR since years. It's still my go-to sim when I want to play IFR, specially now in version 9 with the GTN 750, but ground handling / physics in ELITE are such a "joke" that I smile when I use it, and it becomes apparent how much much better XP and even ESP are... But up in the air, I can fly it's aircraft exactly by the numbers, have a feel of control that is somehow simplified to be usable on a desktop sim but accurate enough to portray the main characteristics of each aircraft model, and it's systems... In X-Plane, I found that the MDLAG 109s specially after their latest ( not more than 1 month ago I guess ) update got really interesting ground physics. The Grumman AA-5 from VFlyteair is very consistent too... I believe that well desgined models for XP11 can bring the limitations imposed by the core physics model to an acceptable level.

[Griphos 138]

@MarioDonick: I wasn't including you in my suggestion that some are arguing just for the sake of arguing.  Perhaps I should have been more specific.  Sorry.  

The use of the term 'simulator' gets pretty slippery on the forums.  People change its meaning to suit their arguments.  Sometimes it means only what we would call a certified training simulator, if that's what they need it to mean.  Sometimes it's used in a much looser sense.  Since the developers of Xplane and FSX and P3D and AF2, etc, use the term "flight simulator" to market their products, and people buy them to use them as flight simulators (for entertainment purposes), then I think the term is appropriate.  But, as with any "simulator" (train, ship, racing, etc), you want the software to do as good a job as possible at representing the real world activity (unless you're just interested in arcade entertainment).  Xplane certainly aims at higher than arcade status, but, with respect to ground handling, they get worse than that.  It's not dumbed down.  It's broken.  

My emphasis on the term was merely to point out that if the end result is totally unrealistic behavior, then the underlying "accuracy" of the code model is immaterial.  Simulators need to simulate well, or what is the point of them, even for entertainment?  

I don't expect hyper-realism from a 2D virtual flight environment.  I do expect that the virtual aircraft somewhat resemble the behavior of the real thing (which, with respect to ground handling of taildraggers, those in Xplane DO NOT).  

@jcomm: I actually have the VFlyte Grumman.  I haven't really flown it much at all (I usually pull out their Cherokee when I pick one of the VFlyte birds), but your repeated praise of it has prompted me to pull it out of the hangar and give it a whirl.  I may look at the 109 as well.  I rarely buy warbirds for civilian flight sims (seems odd just to fly them and not shoot at anything to me), but the Classics Hangar Me 108 Taifun was a favorite when I still flew FSX.  As may be obvious from my choice of real world aircraft to fly, I am a big fan of 1940's airplanes!  :-)

[jcomm 5,930]

In the AA-5 you have the option to set the nose wheel free as in the real aircraft, or simplified ( tied to the rudder pedals ) ...

The 109 is nice, but is has it's problems too... Not on the OP theme, but would call your attention to a weird pitching oscillation that I believe to be tied to the way the slats are implemented in X-Plane.

Austin also informed me that in it's present state asymmetric slat deployment is not possible in X-Plane.

Anyway, I did find the model very well implemented and a joy to use.

 

[olderndirt 2,295]

Curiosity forces me to ask.  On several forums where this directional control problem is featured, different "planemaker" solutions have been offered and a few have been borderline excellent.  My question is if the average simmer can investigate and come close so solving these problems, what are developers doing other than posturing defensively.?  Given, LR has a bug but why are some models infected more than others?  It seems that what's working for some might also work for the others.  Something to cling to 'til LR gets their gyros uncaged.

[jcomm 5,930]

8 hours ago, olderndirt said:

Curiosity forces me to ask.  On several forums where this directional control problem is featured, different "planemaker" solutions have been offered and a few have been borderline excellent.  My question is if the average simmer can investigate and come close so solving these problems, what are developers doing other than posturing defensively.?  Given, LR has a bug but why are some models infected more than others?  It seems that what's working for some might also work for the others.  Something to cling to 'til LR gets their gyros uncaged.

It can even be entertaining to pick a given aircraft model and open it in PM and try to tune down the problem using simply the Artificial Stability section, by making the SAS dance on the rudder for you up to, say, Vs0.

This is the first approach I took long ago in XP10.

You can also open a model that is working "miraculously" well and try to find what's causing the "success"... The Corsair F4-U XP11 available from a well known store is such an example. I was perplexed by it's ground handling and went searching in it's ACF using PM, just to find the author uses an hidden "skate" in the middle of the fuselage, creating the necessary drag and additional longitudinal / directional stability on ground :-) 

I should say that, IRL, windy days are always a "fun" at the airfield, depending on the landing stripe... At LPMN from where I operate the most at the present we have a very thin asphalt rw ( 03-21 ) and taking off with more intense x-wind, and specially landing for instance the Pw-6U glider is a true "nightmare" because even if I use full rudder downwind while I still have rudder authority on the landing run, I know I will still end up exiting the rw highwind side :-)  - X-Plane always comes to my mind after such landings :-)

 

Edited March 12, 2018 by jcomm