Posts by aenbacka

  • Carbon Cub

    Quote from Jet-Pack (IPACS)

    In the video the reason for the crash reset is a propeller collision with the ground. The gear was probably too soft to stop the velocity and deflected enough for the prop to hit the ground.

    Regarding the airfoils: If I recall correctly the Carbon Cub tmd is based on Krzysk's Piper Cub and I already set up the airfoils for that. So the airfoils should be all good in theory. You can of course adjust it for flaps lift and moment coefficients, etc.

    I guess touchdown speed impacts this too. So it is the multijoint Kps that might need to be adjusted?

  • Carbon Cub

    I think the landing speed can be a huge impact here, touching down with 90-100 knots will probably break something. I have some issue getting the throttle reduced.

  • Carbon Cub

    Quote from jayeye2011

    If you didn't have to consider some kind of divide by zero error from improper coordinate values, I would normally try to have a softer shock absorber like in cars rather than a stiffer one if you want a less bouncy landing. But I don't know if that is more influenced by the Kp (spring constant) term or if it is the Kd (damper) term.

    Yes I have been testing with multiple Kp/Kd values for the joints. Now front gears uses Kp around 150000 and Kd 500.

  • Carbon Cub

    Yes it has been quite some investigation to figure out everything, Jet-Pack (IPACS) support has been invaluable. I have the original model in Blender, and I have tried to verify the different R0s for the gears and multibody against this (and as you said, they are relative to model origin).

  • Carbon Cub

    Quote from jayeye2011

    I am not a real life pilot so I don't know how the aircraft is supposed to fly in terms of its takeoff and landing pitch attitude characteristics and how easy it is to be in a nose up attitude when landing. I know that Kaniewski's PA-11 Cub Special is configured so it's natural flight pitch attitude is nose up (not sure if that is real-life accurate) so that it is easier to make a 3 point landing. In contrast, Kaniewski's Stearman Model 75 Kaydet is currently setup more for a nose down attitude characteristic making a 3 point tough to achieve so you end up easily doing 2 point bouncing ball landings which if done incorrectly resulting in a nose down propeller hits the tarmac landing.

    Well it could be the prop that hit tarmac as well, this was most for checking the gears stuff. Not much of a proper landing and touchdown. With the panel it might be easier to see the different values at touchdown time. At the moment the plane has no sounds activated at all.

  • Carbon Cub

    Quote from jayeye2011

    Is your cockpit instrumentation actually working as its hard to tell from the video? If not display the inflight data display ribbon ("I" shortcut key) so we can see what your vertical velocity is on touch down. Also with the VVI from the data ribbon display enabled, using a chase view for landing that gives a view of the landing gear would show (if the landing gear animation graphics are already setup) if it is collapsing or not. For conventional (taildragger) landing gear doing a two point touchdown (two front wheels contact first) is normally prone to a couple of bounces if you are descending too fast, normally a three point landing (both front wheels and tail wheel contact simultaneously) is less likely to bounce. Also, using a chase camera view allows you to see how the landing gear shock absorber is working as long as your gear graphics is working correctly.

    I can make another video with a chase view and data ribbon on. The cockpit is not yet implemented, this is the original one still.

  • Carbon Cub

    I did some quick tests with the gear by editing the ContactSpheres in the tmc and trying to drop it down on the runway. With the Kp/Kd values for the joints for which the plane stands steady on the ground it seems to reset simulation when touching down (it jumps up and down a few times first, as seen in the vide below). How to see what is the cause for the reset?

    Gear testing

  • Carbon Cub

    Quote from jayeye2011

    Wonder if there is an easy way to test how well your gear Kp and Kd setup handles shock absorption, like some kind of drop test from a small preset height above the ground.

    Good question, I guess when we get it up in the air it will be possible to test landings on different types of surfaces. These parameters are difficult to know ideal values before proper testing has been done. First challenge was to get the ground contact to work, so the rest can easier be tested.

  • Carbon Cub

    Jet-Pack (IPACS) Should the value for R0 be the same for both the rigidbody and wheel elements, e.g. as in the following example. I also found in the tmd file an extra multibody_joint element which pointed to an entirely bogus location, maybe that could have caused some of the issues?

    <[rigidbody][RightGear]

    <[tmvector3d][InertiaLength][ 0.420624 0.2 0.8 ]>

    <[tmvector3d][R0][ 0.3046 -0.3654 -0.646 ]>

    <[tmmatrix3d][B0][ 1.0 0.0 0.0 0.0 0.70711 -0.70711 0.0 0.70711 0.70711 ]>

    >

    <[wheel][RightWheelHull][]

    <[string8][Body][RightGear]>

    <[tmvector3d][R0][ 0.3046 -0.3654 -0.646 ]>

  • Carbon Cub

    What is the preferred process for determining / estimating the Kp/Kd values for different elements (e.g. multibody joints, wheels etc)?

    Are there any tools that can display the different forces acting on an aircraft in Aerofly FS4?

  • Carbon Cub

    Quote from Jet-Pack (IPACS)

    There are no collision boxes in Aerofly FS. All collisions are computed based on the actual 3D mesh of the airplane + manually TMD defined wheels.

    The flips or glitches come from an ill-defined joint. Check the joint multis and reduce their Kd and Kp values. You can make then rigid for now to get away from the issue temporarily, with input lock 1.0:

    Code
                <[multibody_joint][JointFuselagePassengers][]
                <[string8][Body0][Fuselage]>
                <[string8][Body1][Passengers]>
                <[tmvector3d][X0][0.0 0.0 1.0]>
                <[tmvector3d][R0][0.0 0.0 0.0]>
                <[float64][Kp][0.0]>
                <[float64][Kd][0.0]>
                <[uint32][Type][1]>
                <[string8][InputLock][1.0]>
            >

    Thanks for the help, it seems to be related to the multibody joints for the gear. The values was 150000 for Kp and 500 for Kd (front wheels), and 10000 / 400 for the tailgear (and these values cause the model to jump around). InputLock stabilizes the system (and the wheels are on the ground, some minor adjustments might be needed), but need to try with some different combinations of values.

  • Carbon Cub

    The aircraft immediately jumps up and flips around when starting the simulation.