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

[ferram4]

@MR4Y: First, are you aware that in Map view you can click on the little button at the bottom of the screen to bring up the navball and then you'll be able to control your vehicle while looking at the orbital Map?

My second thought is that the vast majority of problems you have seem to stem from turning on SAS, particularly when you still have the liquid boosters activated (seriously, get rid of those fins, they're making control on the roll and pitch axes difficult). Perhaps without fins SAS won't cause control issues on that rocket.

Your launch that started at ~7:00 was fairly well done, I'd only advise beginning the gravity turn sooner so that you have a slightly flatter trajectory. That launch was almost exactly what you want to do, though I think you were a little aggressive trying to turn the velocity vector at ~10km. If you had started your gravity turn a second after you lifted off you probably would have had a much smoother launch.

Other suggestions:

Don't worry about roll orientation, just roll rate; roll orientation doesn't matter much on rockets, but roll rate can interfere with control inputs.

Don't worry too much about the velocity vector's direction after the first bit of the gravity turn; just aim prograde and burn along your velocity vector after you've started picking up speed. This will make controlling larger launch vehicles easier when you get to them and helps reduce dV steering loses.

@Silenvio: Good to hear. I figured it was something out of my control, since you were the only person having that particular problem.

[asmi]

I would love some guide how this works, as well as some more details on the in-game functions. I sort of messed about until now, but I am quite sure I am missing quite a bit of functionality.

The best course of action at first is to replicate some real-life vehicle. Then once you get this done and flying, you can start experimenting.

ferram, I wonder if it would be possible to replicate this with FAR:

This thing flies like its designers somehow changed laws of physics! I love this!

[MR4Y]

@MR4Y: First, are you aware that in Map view you can click on the little button at the bottom of the screen to bring up the navball and then you'll be able to control your vehicle while looking at the orbital Map?

My second thought is that the vast majority of problems you have seem to stem from turning on SAS, particularly when you still have the liquid boosters activated (seriously, get rid of those fins, they're making control on the roll and pitch axes difficult). Perhaps without fins SAS won't cause control issues on that rocket.

Your launch that started at ~7:00 was fairly well done, I'd only advise beginning the gravity turn sooner so that you have a slightly flatter trajectory. That launch was almost exactly what you want to do, though I think you were a little aggressive trying to turn the velocity vector at ~10km. If you had started your gravity turn a second after you lifted off you probably would have had a much smoother launch.

Other suggestions:

Don't worry about roll orientation, just roll rate; roll orientation doesn't matter much on rockets, but roll rate can interfere with control inputs.

Don't worry too much about the velocity vector's direction after the first bit of the gravity turn; just aim prograde and burn along your velocity vector after you've started picking up speed. This will make controlling larger launch vehicles easier when you get to them and helps reduce dV steering loses.

Yes, I'm aware, but since the rocket has a tendendcy to overshoot it's trajectory, it's faster to just press m and switch.

the problem with the gravity turn is that there is no middle grounds. If I gently tap the rocket to yaw, it won't do much. If I hold it a little longer, it overshoots the trajectory. If you look at the second launch, I almost crashed the rocket. It doesn't look like much of an issue in video, but controlling that spin and putting the rocket facing up was really hard. The last launch I should had turned earlier, so the apoapsis didn't ended at 200 Km.

[ferram4]

@asmi: No reason you can't make something that maneuverable; the main problem is that it'll be statically unstable, and so you'll have to spend a lot of time keeping it on course.

@MR4Y: Why don't you ever put engines that can gimbal on your rockets? I've only ever seen you use LV-30s, not LV-T45s. That might make controlling it easier. Otherwise, your "there's no middle ground" is pilot error, not design error.

[MR4Y]

@MR4Y: Why don't you ever put engines that can gimbal on your rockets? I've only ever seen you use LV-30s, not LV-T45s. That might make controlling it easier. Otherwise, your "there's no middle ground" is pilot error, not design error.

All engines except for the lander stage are LV-T45's(lander stage uses one Lv-909)

[ferram4]

Hunh... then it's still pilot error; you have everything you need to fly it, you just need more practice.

[MR4Y]

Hunh... then it's still pilot error; you have everything you need to fly it, you just need more practice.

So far, I removed the fins and placed the tanks closer using the small low profile radial decouplers. Now I can use ASAS to hold the rocket in position as I do the gravity turn.

Now to find the problem with ASAS and planes.

[ferram4]

The problem is that control surfaces take time to move with FAR, as opposed to their magic instant deploy without FAR. ASAS doesn't account for the delay. Try going into the ASAS part.cfgs and reducing the kp, ki and kd values. ASAS is a PID Controller, and if it isn't tuned properly your vehicle will be difficult to control with it.

[localSol]

In case anybody missed it, I just wanted to suggest to any new guys or anyone curious about FAR, you can check out the new README.txt included in the FAR 0.95+ archive. Good tips/instructions on how to build and fly with rockets and spaceplanes.

[MR4Y]

The problem is that control surfaces take time to move with FAR, as opposed to their magic instant deploy without FAR. ASAS doesn't account for the delay. Try going into the ASAS part.cfgs and reducing the kp, ki and kd values. ASAS is a PID Controller, and if it isn't tuned properly your vehicle will be difficult to control with it.

Actually, with better construction, true gravity turns and FAR, that rocket can reach Minmus with 30%-40% fuel on the second stage remaining. ATM I'm preparing to circularize the Minmus orbit and still have about 20% fuel on the second stage.

As for tweaking the ASAS values, I have no idea how to do it or even how much I should alter those values.

[asmi]

So far, I removed the fins and placed the tanks closer using the small low profile radial decouplers. Now I can use ASAS to hold the rocket in position as I do the gravity turn.

Now to find the problem with ASAS and planes.

I use MechJeb's Smart ASS in surface mode to control vehicle during ascent as it's may more precise than I ever will be manually. You can try doing the same.

[Van Disaster]

Hmm, the drag changes might explain why my rockets have been tumbling on their usual ascent profiles too, I tend to use the KW extended fairings by default & the CoD is going to be way up in the nose - which is ok until you start going sideways. Other than use a straighter profile which is what I've resorted to, I'm not sure what else to do about that. I doubt going slower is really going to help that much, and it'd take fins the size of wings to provide a balancing moment. This is where having 70km of atmosphere hurts, not enough time to fly an ideal path.

Most definitely need a blended wing large fuselage now. Something that looks like a giant Manta Ray perhaps.

[m4ti140]

Ferram, are you planning to add heatshield aerodynamics (like lift at non-retrograde position of pod)?

[ialdabaoth]

ferram:

I just released a new version of ModuleManager, which fixes several bugs. Chief among them, overloading @module = Part works correctly now!

[ferram4]

@MR4Y: The link I gave for PID controllers has some info on what all the numbers mean. In short, increasing Kp and Ki will make the system respond faster, but make it less stable (read: oscillate more) and increasing Kd will cause the system to try to limit motion and make it more stable; a proper combination allows the system to respond very quickly, but without oscillations.

@Van Disaster: Make sure that the KW Rocketry parts have the proper size attach nodes; the engines might not be making the drag that they should, making the rocket less stable.

@m4ti140: The command pods already make some lift when they're oriented off of retrograde, just like all the other parts do.

@ialdabaoth: Noted, I'll test it out and release a new version of FAR with the modified ModuleManager.

[MR4Y]

@MR4Y: The link I gave for PID controllers has some info on what all the numbers mean. In short, increasing Kp and Ki will make the system respond faster, but make it less stable (read: oscillate more) and increasing Kd will cause the system to try to limit motion and make it more stable; a proper combination allows the system to respond very quickly, but without oscillations.

I'll probably test it along with Modular Fuel Tanks as well, since it now allows you to change ASAS PID values.

[Wert]

Anyone know why the Command Pod Mk 1 being forced to face sideways by aerodynamics now instead of heat shield-first? Since I have deadly reentry, this is bad news for the kerbals, although when reentering from a 100km orbit one actually survived.Here's an image:

http://i.imgur.com/Dc0pWMR.png

Thank you to anyone who can explain this.

Edited June 16, 2013 by Wert

[longhornchris]

I'm using MechJeb 2.08 beta. Has anybody else had any luck getting a good starting point for MechJeb's Attitude control PID? Adjusting the PID values work with FAR should be doable, but I'm having trouble getting there.

[ferram4]

@Wert: The reason is that unused attach nodes add drag, and in the latest update I also updated the location of the center of drag for each part based on where the unused attach nodes were. Since the "heat shield" is where an attach node is, the CoD gets shifted towards that. I did add a small correction to make the pods stable, but I apparently didn't test enough. It will be fixed in the next release.

[velusip]

Getting log spam. Over two-hundred thousand lines of:

NullReferenceException
  at (wrapper managed-to-native) UnityEngine.Component:GetComponent (System.Type)
  at UnityEngine.Component.GetComponent[FARWingAerodynamicModel] () [0x00000] in <filename unknown>:0
  at ferram4.FAREditorGUI.Update () [0x00000] in <filename unknown>:0
  at ferram4.FAREditorGUI.ControlSurfaceAttributionGUI () [0x00000] in <filename unknown>:0
  at RenderingManager.OnGUI () [0x00000] in <filename unknown>:0

NullReferenceException
  at (wrapper managed-to-native) UnityEngine.Component:InternalGetTransform ()
  at UnityEngine.Component.get_transform () [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.recurseParts (.Part part, UnityEngine.Vector3& CoM, System.Single& m) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.findCenterOfMass (.Part root) [0x00000] in <filename unknown>:0
  at EditorMarker_CoM.UpdatePosition () [0x00000] in <filename unknown>:0
  at EditorMarker.Update () [0x00000] in <filename unknown>:0

NullReferenceException
  at (wrapper managed-to-native) UnityEngine.Component:SendMessage (string,object,UnityEngine.SendMessageOptions)
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.recurseParts (.Part part, UnityEngine.Vector3& CoL, UnityEngine.Vector3& DoL, System.Single& t) [0x00000] in <filename unknown>:0
  at EditorMarker_CoL.Update () [0x00000] in <filename unknown>:0

NullReferenceException
  at (wrapper managed-to-native) UnityEngine.Component:GetComponent (System.Type)
  at UnityEngine.Component.GetComponent[FARWingAerodynamicModel] () [0x00000] in <filename unknown>:0
  at ferram4.FAREditorGUI.Update () [0x00000] in <filename unknown>:0
  at ferram4.FAREditorGUI.ControlSurfaceAttributionGUI () [0x00000] in <filename unknown>:0
  at RenderingManager.OnGUI () [0x00000] in <filename unknown>:0

I'll do some triage later if necessary (removing all other plugins and maybe replicating this using a simpler craft files).

[ferram4]

@velusip: If you manage to replicate it, would you mind posting the entire output_log and the craft file that caused it with full instructions on how to cause it?

[m4ti140]

@m4ti140: The command pods already make some lift when they're oriented off of retrograde, just like all the other parts do.

Thanks for info! That's good to know. I have a question about wing parameters: How to handle a part that is an actual wing, not a canard or such, that has built in control surfaces? Should I add both FARControllableSurface and FARWingAerodynamicModel or just FARControllableSurface?

[nmdanny]

I have some questions regarding how the mod works:

1. What is the purpose of setting the control surface's axis? I can understand assigning them to brake or flaps, but I have every control surface for pitch, yaw and roll and it doesn't seem to have negative effects on my craft.

2. Is there a guide for the advanced portions of this mod, such as what values to use in the flight assistance system, or what do the graphs and terms in the analysis panel mean. (A short guide on checking if my craft is stable and flyworthy?)

[m4ti140]

I have some questions regarding how the mod works:

1. It does affect your a/c - if you make a standard-configuration plane (main wing and rear stabilizers), if you leave pitch at ailerons, they will counter elevator's input - if you pitch up they will act like spoilers, pushing you back down, and if you pitch down, they will act like flaps, lifting you up. Also if you leave roll on rudder, it may overdo its input, leading to unnecessary yaw + IRL it's pilot's job to take care of rudder and apply correct yaw to prevent side-drift. Leaving pitch on rudder and yaw on elevators/ailerons shouldn't do any harm since they don't work in those axis anyway but only if your wings/horizontal stabilizers are perfectly horziontal, and vertical stabilizers are perfectly vertical, otherwise they will make unnecessary input in axis they are not meant to work in. Also leaving roll on elevators is only a good idea if you build a fighter and use winglets for elevators.

2. I guess some study on aerospace engineering XD. I'm starting my studies in October, so I guess I will find out . But wikipedia should be enough.

@Ferram - I think I also hit the node drag problem, I was reentering with soyuz descend module from bobcat - since it has both, top and bottom node, free at reentry, the torque from drag they made them cancelled out (though it should've been applied on top node only, since it isn't aerodynamical, forcing proper orientation) which togeather with intentionally offset CoM allowed me to stay in correct orientation. However, the drag of whole pod was so ridiulously high, that while normally I would probably overshoot the target with FAR (since it lowers drag), I landed almost 300 km short of target. I had like 5G on flattest possible reentry trajectory and didn't even see reentry plasma. It worked almost like a parachute.

Maybe you could allow to disable additional drag at specific node with a part.cfg parameter/module (name = FARNodeDrag; disableDrag = [node name])?

Edited June 16, 2013 by m4ti140

[Doomydoom]

The best course of action at first is to replicate some real-life vehicle. Then once you get this done and flying, you can start experimenting.

ferram, I wonder if it would be possible to replicate this with FAR:

This thing flies like its designers somehow changed laws of physics! I love this!

I made the SU-27 in KSP 0.19. I ended up having to make a small plugin to automatically control my leading edge slats, and I also had an engine with 15 degrees of gimbal range. It was unstable but manageable, could do the Cobra and Supercobra maneuvers.