[1.3.1] Ferram Aerospace Research: v0.15.9.1 "Liepmann" 4/2/18

[randman22222]

randman2222, as for you, you need to provide reproduction steps and logs. I haven never been able to cause any issues like that in the current build, which means that something you're doing is different so you have to tell me or else there's nothing I can do. I think you just installed it wrong, tbh.

Okay; I just retried it with von Helmholtz. Steps are as follows, with screenshots here and logs here:

-Fresh Steam install, after deleting leftover folders from uninstalling.

-Downloaded von Helmholtz from Kerbal Stuff, and merged the GameData folder.

-Launched the launcher, turned off 64bit.

-Launched KSP, only settings changed were keyboard controls (I have a Dvorak keyboard).

-Created a sandbox save with default settings.

-Recreated a ridiculous awesome plane I saw on imgur earlier today.

-Ran FAR calculations, starting with AoA and mach sweep, then stability derivatives.

-Launched plane; after the engine didn't start, I did the quicksave/load trick to start it, but the intake was occluded.

-Quit.

I hope this is what you need? If not, I'll try to help however I can!

[ferram4]

@VanDisaster: I don't change how the stock airbrakes behave. At all.

@Boris-Barboris: Oh, yes, I originally added the exponential decay code, DaMichel just adds comments everywhere when he contributes anything (and is much better than me in that regard). The original reasoning was that it not only closely follows how servos respond to deflection commands but also makes for a very easy-to-handle result for control systems (as I'm sure you're aware). So the first-order-system response / exponential decay isn't going to go away.

That said, depending on how the overhaul works out the current methods you're using to figure out what kind of control FAR can provide might not work. Once development of that feature starts in earnest it'll show up on the FAR github on a more recent wing overhaul branch; feel free to mess with the builds from that (I always upload the latest dll) and either comment there or PM me here if you need anything on my end to make hooking into that easy.

@randman22222: Well, the first thing I need is the craft file. Preferably with all mods removed. If that's not possible, stripped to the barest minimum of mods necessary to cause the issue.

And then I need a full, complete copy of your output_log.txt (or Player.log if you're on Linux / Mac) so I can see what when wrong. And I mean complete; don't snip anything, lest something important get lost in that.

[Boris-Barboris]

Yep, it makes control surface function differentiable and have linear first order representation, I was surprised when analyzed it and was wondering if you intentionally did it. Thank you for that, though I had to control stock planes, so in the end that benefit was in vain: now I have ugly branching for FAR and stock models. I don't reference FAR dll, I simply assume that FAR uses default 0.25 csurf factor and track control surface position from user input, mimicing FAR actuator code. It works, and if you won't change general exponent idea and default factor, everything will keep working.

Edited November 13, 2015 by Boris-Barboris

[ferram4]

Actually, I think the stock control surfaces do model it as a first-order system. I'm not sure exactly how the actuatorSpeed field is used, but I think that they do end up modelling an exponential decay towards the desired control deflection.

A simple test would be to load up the game without FAR and one of the control surfaces modded so that it has an actuatorSpeed field with a very low number and then see how the control surface behaves.

[Sean Mirrsen]

With the advent of the new version I felt like trying FAR again, and of course first thing I did was try it on the most unlikely to work correctly craft I could pick, the recreation of my old "Space Harrier" SSTO from way back in 18.1:

Surprisingly, the test run proved successful, for a given, very generous definition of "success". The craft actually made it to space, and back down to earth, gently touching down for a landing, where it was swiftly consumed by a sudden bump in the terrain.

The test, however, had some results that made me want to ask a few questions.

1. What arcane ritual must be performed to make the flap/spoiler functionality work on control surfaces? The switch to FAR completely shot all the control surface setups I had on the craft (which I found out only already in flight), but I made do until I had to land, whereupon I could not make the tail elevators trim to compensate for the shift in CoM, as I could do in stock KSP, which made the flight unnecessarily loopy.

2. Maybe it's just the sort of designs I make, but whatever version I try to "try FAR again" with, I keep getting weird asymmetric forces acting on the plane. In this case it kept pulling and rolling left, even with SAS trying to compensate. Sideslip issues is one thing, the weird (to put it mildly) M-shaped vertical stabilizer and rudder can and did cause a few control problems, but those were easily identified as such. A constant pull to one side, however, is as annoying as it is mysterious.

It wouldn't surprise me if these were entirely my fault, but I have to ask just in case it's a known issue or a possible fixable bug of some kind. Craft is completely stock, besides whatever changes FAR itself makes.

[Boris-Barboris]

ferram, stock is

float max_delta = TimeWarp.fixedDeltaTime * actuator_speed;
return prev + Common.Clampf(desire - prev, max_delta);

[ferram4]

@Sean Mirrsen: 1. Trim settings with FAR are exactly the same as trim settings in stock KSP; ALT + WASDQE. If you're intending to use the flap options for trim, then you need to assign the controls to respond to flap action groups in the editor (or toggle using the more/less deflect buttons in flight), and then tweak the flap deflection slider to get the deflection you want. For elevators used to trim the plane up, you'd want to set that number to be negative.

2. That's normally a sign of uneven flexing on the plane. Should also happen in stock, but now that I think about it, since FAR actually cares about sweep angle, flexing could reduce / increase the sweep on one wing, increasing / decreasing the lift it produces and causing a roll tendency.

However, if all your planes have massive anhedral wings like that, they're going to want to roll all over the place no matter what you do. Anhedral wings are less roll stable regardless of the flight model, FAR just might make the effects more noticeable. Oh, wait. I see. Swept wings + not much vertical tail. Yeah, the problem with that design is that it has no yaw stability and would like to slide on sideways.

So I think most of these are your fault. That said, if you do find anything funky, issue reports are always welcome so long as they come with repro steps, craft files (if the repro steps include a specific craft), logs, etc.

@Boris-Barboris: Hrm. Damn. I brought up changing it to exponential decay since that would be much better for you, Sarbian and the other MJ guys, and for Squad if they decide to overhaul SAS. We'll have to see if anything pans out though.

Edited November 13, 2015 by ferram4

[Boris-Barboris]

I don't care, it's not so easy anyways with both cases.

[Sean Mirrsen]

1. Trim settings with FAR are exactly the same as trim settings in stock KSP; ALT + WASDQE. If you're intending to use the flap options for trim, then you need to assign the controls to respond to flap action groups in the editor (or toggle using the more/less deflect buttons in flight), and then tweak the flap deflection slider to get the deflection you want. For elevators used to trim the plane up, you'd want to set that number to be negative.

Aha. Okay, apparently I needed to click on "more" several times to actually get the angle I select. And it's slow. Hmm.

You know, maybe it's not a bug, but I think it's unwanted behavior. The speed at which the surface adjusts into the selected deflection angle (or rather, the selected percentage of the selected deflection angle) is set based on the selected deflection angle. This creates a situation where you can't adjust the selection angle from a large number to a small number, or especially to zero, without making the control surface essentially stuck. I.e. if you set the deflection to 30 degrees and click "deflect more" twice, then select 5 degrees, the surface will track towards the five-degree deflection at a glacial pace. Not sure how you would effectively fix that, but it's a thing I just found poking around the interface, so reporting.

2. That's normally a sign of uneven flexing on the plane. Should also happen in stock, but now that I think about it, since FAR actually cares about sweep angle, flexing could reduce / increase the sweep on one wing, increasing / decreasing the lift it produces and causing a roll tendency.

However, if all your planes have massive anhedral wings like that, they're going to want to roll all over the place no matter what you do. Anhedral wings are less roll stable regardless of the flight model, FAR just might make the effects more noticeable. Oh, wait. I see. Swept wings + not much vertical tail. Yeah, the problem with that design is that it has no yaw stability and would like to slide on sideways.

No, see, lack of stability I understand. It's the preference in instability that I find odd. I.e. looking at the aerodynamics overlay (I don't know if it's adjusted for FAR or not), there is a noticeably increased drag on the left side of the craft, which persists regardless of which side of the craft leans into the airstream. Among other things, from the two spokes in the middle over the aerospikes, it looks like the swept wing on the left gets more drag than the one on the right, even as the plane is slightly veering left, which makes no sense to my feeble understanding of aerodynamics. Isn't a swept-back wing supposed to get less drag as it is swept further back, and more as it is swept forward - i.e. a "shuttlecock" effect?

I think I'll just toss in the craft file as well, so that you can try it out if you want and maybe spot what I'm missing here.

(If not, well... it's a pretty craft, at least I think so. Maybe other people would like to fly it around. )

[Doke]

I was experiencing the exact same roll instability that Sean Mirrsen describes. I just installed von Helmholtz, and that has fixed it completely.

It was very strange, though, effecting even craft in near-vacuum, capsules floating on parachutes at 4m/s, etc. Every time I deactivated SAS, there it was.

[cantab]

I was experiencing the exact same roll instability that Sean Mirrsen describes. I just installed von Helmholtz, and that has fixed it completely.

It was very strange, though, effecting even craft in near-vacuum, capsules floating on parachutes at 4m/s, etc. Every time I deactivated SAS, there it was.

That sounds suspiciously like leaving the wing leveller on. That's confused the heck out of me SO many times.

[ferram4]

Well, I tried the craft, and I can't reproduce any issues. It's yaw happy, yes. But not roll happy. Aerodynamic forces appear to be placed equally from everything I can see. It flies like a slightly yaw unstable brick, which is kinda what I'd expect. Not sure what your exact issue is then.

[Doke]

That sounds suspiciously like leaving the wing leveller on. That's confused the heck out of me SO many times.

I had no idea that could happen. I feel really dumb now.

[Sean Mirrsen]

Well, I tried the craft, and I can't reproduce any issues. It's yaw happy, yes. But not roll happy. Aerodynamic forces appear to be placed equally from everything I can see. It flies like a slightly yaw unstable brick, which is kinda what I'd expect. Not sure what your exact issue is then.

Hmm. I think I can chalk that one up to a lot of tiny issues manifesting as one bigger one. I'm playing on a tablet, so my framerate is low with this many parts, which might lead to the flight integrator accumulating small errors. The control surfaces were not set properly, leading to some of them potentially acting against the intended controls. And the craft is aerodynamically unstable, making SAS want to correct it often, and causing lots of little movements from the poorly-setup control surfaces, which might have made the struggling physics engine show an anomaly.

I any case, I appear to have fixed the problem by setting up the avionics as they were intended, and changing the shape of the rudder to provide more authority and stability.

I should note that the craft is even easier to SSTO in FAR than in stock, despite having this odd aerodynamic makeup and not being a "rocket with wings" or Rapier-powered sort of SSTO. Curious.

[iospace]

Found yet another bug with flap/spoilers: default deflection is 15, but if you change it, it's next to impossible to set it to 15 again, or well, I have yet to reset it to that.

[Romby]

Hello. I'm having problems with the settings in Space Center Menu. Every time I change something it gets reset after one flight. Have tried changing flap buttons and turning of aero stress. Is there a settings file in GameData folder i can change to keep my new settings (couldn't find it myself)?

[PickledTripod]

Seeing as this thread has over 11,000 replies I'm not going to read though all of it, but is there a compatibility issue with the Trajectories mod in 1.0.5? If I switch to map view in atmosphere with a shuttle or SSTOs the wings just explode, from aerodynamic stress according to F3. It does that even when the vessel is sitting on the pad, at 0 m/s.

There are a few messages in the debug log that seem relevent:

[Log]: Trajectories: Initializing aerodynamic model...
[Log]: Updating vessel voxel for Procyon Prototype
[Log]: Error in mesh triangle; triangle plane components are NaN or triangle is degenerate; FAR unable to use this triangle
Error in mesh triangle; triangle plane components are NaN or triangle is degenerate; FAR unable to use this triangle
[Log]: Std dev for smoothing: 3 voxel total vol: 1377.48897996264 filled vol: 654.651918644401

[tetryds]

@Sean Mirrsen: Good to see that you solved most of you problems.

But yeah, FAR is a heavily used mod, and its issues are taken seriously, you will hardly ever find a serious bug that is caused by it.

Now if you want to continue discussing about your design please move to the FAR craft repository.

@iospace: That is not a FAR issue, it's about how the stock tweakables are set.

You won't even notice the difference if you select any value close to 15 though.

[randman22222]

@randman22222: Well, the first thing I need is the craft file. Preferably with all mods removed. If that's not possible, stripped to the barest minimum of mods necessary to cause the issue.

And then I need a full, complete copy of your output_log.txt (or Player.log if you're on Linux / Mac) so I can see what when wrong. And I mean complete; don't snip anything, lest something important get lost in that.

Hey, here are the output_log.txt and the .craft file:

https://www.dropbox.com/s/8fcmwebe8wecenj/output_log.txt?dl=0

https://www.dropbox.com/s/lm3dwx7c5d2yaln/Fire%20The%20Aeromech%20Engineers.craft?dl=0

[Gryphon]

Seeing as this thread has over 11,000 replies I'm not going to read though all of it, but is there a compatibility issue with the Trajectories mod in 1.0.5? If I switch to map view in atmosphere with a shuttle or SSTOs the wings just explode, from aerodynamic stress according to F3. It does that even when the vessel is sitting on the pad, at 0 m/s.

You might ask on the Trajectories thread if you think there is a compatibility issue. FWIW, I don't recall seeing a mention of such an issue in that thread.

With that said, what version of Trajectories (and FAR) are you using? Also, are you using other mod parts? If so, can you reproduce the problem without them?

[Guest]

So what is this new velocity indicator ?

[Crzyrndm]

@ferram4

Adding an IAS/EAS speed control to Pilot Assistant, I was benchmarking the output speeds against those given by FAR. Got it mostly working, but there's still a small discrepancy between the two values. My current assumption is that this is because FAR uses a [URL="https://github.com/ferram4/Ferram-Aerospace-Research/blob/master/FerramAerospaceResearch/FARGUI/FARFlightGUI/AirspeedSettingsGUI.cs#L163"]generic atmospheric density[/URL] rather than the one specific to the vessel (vessel.atmDensity. Different temperatures at different locations now the thermal system is in place). You're the expert here though so is that going to be the case?

[ferram4]

[b]@randman22222: [/b]I cannot reproduce the issue. The vehicle launches fine, its engine turns on fine, and everything seems correct. I suspect your issue lies in another mod interfering with FAR or some other program you have installed preventing FAR's worker threads from running.

[b]@PickledTripod: [/b]None of the log that you have snipped out has anything to do with what you're describing; all of those are lines that occur in other installs without any issue (except perhaps the Trajectories line). Without context they aren't much use anyway, so please, if you're going to upload anything from a log in the future, upload the entire log, not a disconnected snippet.

In any case, those issues should not occur during normal use if Trajectories is set up correctly. If it's calling functions that could cause aerodynamic failures though (and that are only supposed to be called by the main game for the purposes of physics simulation), then that could happen. If that's the case, it needs to be fixed on Trajectories's end, I can't fix that.

[b]@RevanCorona: [/b]It's in the editor. You will know it when you see it, it's kinda hard to miss.

[b]@Crzyrndm: [/b]Could be that. Except FAR's method for reading atm density should just read back the current air density now, so I don't know what could cause the discrepancy.

[Kagame]

I've got a sneaking suspicion that the new structural fuselage isn't voxellizing right, pending further testing. (Downright odd sideslip behavior, vessel far draggier than I expected) As it could simply be an issue with the main wings on the fighter I created being incompatible with FAR and/or 1.0.5, I'll go ahead and do some further testing using a more.... conventional design before setting off any frantic scrabbles through source code. Heck, everything could be just fine, that one fighter might have just been really badly advised.

EDIT: Stock normal-shaped wings check out on an airframe with the new structural fuselage. Disregard this entire message and have a great day. Edited November 17, 2015 by Kagame

[FungusForge]

Newbie to FAR with a question.

Does FAR simulate [URL="https://en.wikipedia.org/wiki/Ground_effect_(aerodynamics)"]Ground Effect[/URL]?