Does JSBSim NOT respect FGs internal properties?

[swampthing]

Bouncing is only is only caused by ground contacts being done wrong not a hard fix. One can argue YASim is always a lawn dart when you crash, to me it doesn't matter. Its certainly not an argument to make against an FDM. I like a nice crash for the pilot if he does so, the ground contacts get some extra attention. some of us have made crashes look so cool people just have to crash a few times and get it out of their systems.

[swampthing]

FYI Helijah I just want you to know that if you look at the post that spawned this one. Yeah you know the one. It wasn't about a competition. JSBsim was chosen out of preference. I'm glad you made the model so it was there to work with but it was a preference on the FDM not a competition for me to show you or anyone else up. I don't get down like that. You eluded to that in this thread and for me that is not the case. My name wasn't mentioned but I just thought you should know.

[bugman]

Note that the default YASim crashing behaviour - a "lawn dart" as you call it - is actually a deliberate and ancient choice by the FlightGear developers. Well, that's what I remember anyway. It has to do with keeping the core of the flight sim non-violent.

Regards,
Edward

[Octal450]

Helijah's position is that a poorly written JSBSim FDM will allow a plane to bounce off the ground, land on water, and do other weird thing.

And in fact, that is still true today. Crank up the springs on the ground reactions and you'll see what I mean.

When you add complexity and configurability, you also increase the amount of parameters that must be carefully adjusted before stable behavior is achieved. You can't just plug stuff into AeromatiC++ for example and expect anything realistic (as that really does a similar job like YAsim, generic tuned via parameters that might even be guesses!, but its a but more better as much more parameters are used). I always go for completely custom FDM using calculations for this reason, but it takes time to get working.

I assume that is a language issue communicating these points with the words "not respecting FG internal properties", i.e. "properties" is used as a common language term rather than referring to the "property tree" which is causing a lot of confusion between all parties. Emmanuel, is this interpretation correct?

Most likely considering it is through Google Translate. JSBsim not respecting the property-tree makes 0 sense whatsoever.

Kind Regards,
Josh

[bugman]

Most likely considering it is through Google Translate. JSBsim not respecting the property-tree makes 0 sense whatsoever.

It's clear that some people have misunderstood this as being related to the property tree. For those knowing the difficulties of non-native speakers, it's clear that that was not the original meaning. And, knowing Emmanuel, I'm also pretty sure that this is just a non-native speaker issue and that he does not need auto-translators.

Regards,
Edward

[Octal450]

When 5H1N0B1 spoke to him about the Mirage, he told me helijah used Google Translate and didn't speak English, but if that is wrong, I apologize.

J

[Thorsten]

And in fact, that is still true today. Crank up the springs on the ground reactions and you'll see what I mean.

Well - that's a feature. JSBSim isn't locked in the assumption that it simulates a plane - you can do a rubber-ball bouncing off the ground with it just as well and that's just as valid a use case - and in this case you require bouncing.

So it's true, if you instruct JSBSim to treat your plane like a rubber ball it will do that - the fault however is not with JSBSim in that case

All I can tell you is to make your own tests because if the data is real, the sensations you will have while flying, the impressions you will feel will be much more important than anything else.

Challenge accepted.

Let's consider your own Space Ship One - I've actually made tests with it, and I'm going to lean out of the window here and tell you that the numbers are clearly wrong - you don't reach anywhere near 100 km altitude with it

The JSBSim X-15 on the other hand does ballistic arcs to realistic top altitudes.

So there's something wrong here - I'm not a YASim expert, but I'm tentatively going to blame the engine simulation which removes power as we ascend... and I challenge you to fix that.

[swampthing]

Note that the default YASim crashing behavior - a "lawn dart" as you call it - is actually a deliberate and ancient choice by the FlightGear developers. Well, that's what I remember anyway. It has to do with keeping the core of the flight sim non-violent.

What do you mean by "non violent"? Can a crash not be simulated? Is that too violent? Helijah YOu can make a good FDM from AeromatiC++ if you have the data and work hard enough at it. There are so many ways to configure things without going to nasal to correct what YASim doesn't give options for. And you argue it will be like YASim well you pretty much said an AeromatiC FDM will be no good and just guesses like YASim. No? Ether way its how much you are willing to put into the FDM. Crap in crap out. Thats what you will get. Of course if you just plug numbers int AeromatiC++ things will not come out just right. It doesn't ask enough questions for that. FCS is always spit out the same but that can be configured. Look at the data you have set the degree or radians that the control surfaces move, use filter to make them move at the realistic speed. The options are nearly endless. Creating properties is good because then you can make you gauges read correctly and for some there are no default properties. You can do the same for YASim in nasll and people do it so what is the problem there? I prefer not to have to go to nasal but in the Jaguar for instance we had to over-ride the default flaps because slat must come out before flap. Sometimes you have to do what must be done for things to be realistic. That was all started with and aeromatiC++ FDM the FCS is pretty much still the same Richard was nice enough to make a great FDM using VSP wihch is not an easy task. I've worked with you Helijah on some of you planes testing them and giving you feedback and you corrected problems in YASim. I was pleased to work with you doing it and it was much better, We should do that more often. YASim is made ina way that I have trouble wrapping my head around though so I go with the other FDM. I've read Bukaroo's page and it just seem to strange to me. The engine power seem to always need to but higher than the real number, Thrust for instance. So its a guessing game trying to get the plane to fly correct at higher altitudes that data says it should. JSBsim big monster turbo-superchargers are maybe much more of a headache though. There are faults.

I don't see what the problem is though. "Not respecting internal properties. So what? There is no problem with improvement. What properties are we talking about that have you so upset? I'm not trying to be mean. I'm trying to understand where you are coming from.

[Octal450]

And in fact, that is still true today. Crank up the springs on the ground reactions and you'll see what I mean.

Well - that's a feature. JSBSim isn't locked in the assumption that it simulates a plane - you can do a rubber-ball bouncing off the ground with it just as well and that's just as valid a use case - and in this case you require bouncing.

So it's true, if you instruct JSBSim to treat your plane like a rubber ball it will do that - the fault however is not with JSBSim in that case

True, I never meant otherwise.
It was never a problem, my argument for when bugman mentions "JSBsim" evolution, this isn't something that evolved or even has to, as its correct (for the most part, always room for improvement!). The MD-11 springs are working really well for me, after some fine tuning and calculations I achieve behavior that is very close to what the real aircraft is doing.

YOu can make a good FDM from AeromatiC++ if you have the data and work hard enough at it.

That's not entirely true. AeromatiC++ also just modifies a generic flight model using input parameters. But it gets closer than YAsim because A, the generic model is better, and B, far more data is used.

I always go with custom calculations and methods, and it knocks the socks off the old AeromatiC (even though I matched the data from the real plane!) according to actual pilots!

Kind Regard,
Josh

[swampthing]

That's not entirely true. AeromatiC++ also just modifies a generic flight model using input parameters. But it gets closer than YAsim because A, the generic model is better, and B, far more data is used.

And you can change anything you want in the FDM it doesn't matter if you start from scratch or start with AeromatiC++ anything can be changed. If your good enough to start from scratch good for you. If you are still amateur and are in the learning process aeromatiC++ is a good place to start, At least you will get something going and not throw your hands up and give up. You've moved beyond that point but don't tell me you never used aeromatiC++. None of us were born knowing everything. But we've had this discussion before. I believe telling people they can never get it right just takes away the incentive to even try. You have to crawl before you run. You may start with AeromatiC++ and later scrap it and make it from scratch. You may start with AeromatiC++ and have hardly anything left of what it originally spit out. Those of us that have worked with it know full well it gives generic outputs. All can be changed later as you find data which can be very hard to find sometimes. So having said that it can be true if you work at it long enough and find the data you need. Either way you can create one from scratch with not enough data and have a hack job or use AeromatiC++ with not all the data it even ask for and have a hack job, Just know that if you enter one thing wrong into aeromatiC++ you are going to have problems. Like when I entered the wrong numbers for tail plane area, it through everything off all the way down the line and cause the death pitch that you have in a lot of YASim planes. Crap in crap out. Look at a lot of planes that use JSBsim in FGaddon that seem pretty poular and you see aeromatic at the top..... Something there may not be perfect but the end user seems to like flying them, its developers that nitpick each other to death. haha

Steve

[Octal450]

Very old thread, I apologize for bumping but I just now saw this and wanted to say one more thing:

AeromatiC++ anything can be changed

True, but once you start modifying parameters it is no longer a AeromatiC++ FDM, as an AeromatiC++ is just a type of automatically generated pre-configured JSBsim FDM. It is based off of it and maybe eventually not even that anymore, but once you start changing things, it starts to become custom. The more you change, the more custom you are.

Kind Regards,
Josh

[S&J]

Custom in reference to aeromatic.

Vanilla jsbsim as per their manual is so old hat and frankly fails to deliver

[Octal450]

Right.

Hence clarifying. I see nothing wrong with using AeromatiC++ for a baseline, then going through and refining it. But on its own it falls flat similarly to yasim as you said.

Kind Regards,
Josh