Solved Unwanted FFB cobblestone effect

Discussion in 'Mod Support' started by fufsgfen, Jul 28, 2018.

  1. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I did even try to increase numrays, but does not effect this, just made brakes smoke while handbrake on.

    You can see from the video how FFB is like driving on cobblestone when I turn.


    It is just AM export, but other AM exports don't have this cobblestone effect when cornering and I have no idea where this would come from.

    I need to figure out if it is loose tie rod ends, is it body jbeam or front suspension jbeam related.

    What typically causes this kind of effect? That would help in narrowing where to look as poking randomly without any idea is probably not working this time.

    Update: Steering damper has zero effect to this.
     
    #1 fufsgfen, Jul 28, 2018
    Last edited: Jul 28, 2018
  2. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    Maybe some beams are too stiff causing vibrations? Maybe some nodes need to be heavier?

    Could be the slidenode hitting the end of the steering rail causing instability?

    It could be a lot of things really.
     
    • Like Like x 1
  3. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    That is what worries me, need somehow find way to limit possible things as it could be too many things.

    Slidenode hitting end of rail it probably is not as that is happening with so little steering input.

    Wheels do get kinda wobbly when shaking happens, but I did see similar wobbly wheels with ETK800, just not as severe.

    Also it seems to be fine on straights and mild cornering.

    However front end is kinda soft, it does flex quite a bit up and down on heavier bumps, but this goes against your thought of too stiff, but could too soft be a reason?

    I might test adding some bracing beams to make front end more solid, maybe that reveals something more.
     

    Attached Files:

  4. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    I haven't checked your car with FFB because its a little late in the day for me to be setting the wheel up etc, so I have just given it a quick flick around with a 360 controller to see if I can spot anything quickly.

    When I said "stiff" I meant the BeamSpring values could be too high, since that can cause vibrations to happen in vehicles. The suspension however is way too soft, and could be a part of your issue if it is causing the suspension to wobble up and down, since the game could interpret suspension vibrations as FFB. Probably a good first thing to try, since if it doesn't fix the problem, at least you now have nice suspension. Copying the suspension spring/damp values from the ETK K would probably be a good starting point.

    It could also be to do with having such a stiff anti-roll bar too. Since it is having to work super hard to keep the car flat with such soft suspension. This would be my best initial punt at an answer. When played in slow motion, the whole anti-roll bar node beam assembly shakes a lot on your vehicle, hence why I think this might be the issue.

    The wheels on the official vehicles do have a small amount of wobble to them, when played in slow motion I am pretty sure yours have more though. Cars in real life can actually have some wobble to them when under heavy loads, if you watch a wrc car in slow motion it can be seen fairly clearly. On my mods I have found it possible to adjust the level of wobble by fiddling with the stiffness of the steering etc. However I doubt that is the issue you are facing.

    The way FFB works in BeamNG is that you specify two nodes per wheel, from memory I think these two nodes are specified on the wheel, inside the "wheel data" section of the jbeam. I think they are the "steerAxisUp" and "steerAxisDown" nodes. The game then uses them to calculate how the FFB should be applied. I must admit that I haven't looked into specifically how it figures it out from those two nodes. But I can say with almost certainty that they won't be an issue, since I highly doubt they would have put the wrong nodes in them.
     
    • Like Like x 1
  5. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Suspension is perfect, certainly not too soft for a gravel rally and going really fast at ECA jumpy streets, it is soft on purpose, on that car looks will be deceiving, that is on purpose too :)

    If you take ARB off, it has no effect on FFB cobblestone effect, it just shakes it.

    I have tested without steering damper, crazy high steering damper that cuts all high frequency over 10Hz and that has zero effect.

    I put more weight on suspension nodes, on tie rods etc. no effect.

    Added some beams to brace front end, but neither that had any effect.

    Playing with beam damping did not have any effect to issue either.

    Maybe it is wheels and tires themselves?

    Or then it is that this specific AM exported beam structure just is not going to work any better without complete revamp of almost everything.

    I highly doubt it being softness of suspension as it has correct damping and it is calculated to be in correct range for it's purpose.

    Hub, tire, suspension arms, something in there is my guess, but haven't found way to affect that.

    Rear wing is bit less than optimal, but haven't got around to make it adjustable yet, so over 200kph is really tricky for now, but this FFB issue is just preventing any progress.
     
  6. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    You could try widening the wheel nodes (the two nodes the wheel is attached to), as it currently stands they are narrower than the wheel rim and tyres. In my experience that can cause some issues with wheel wobble. Since there is a lot of force on them even an extra few cm can sometimes make a big difference, since it makes the whole hub stronger and stiffer. I know on some of the official vehicles like the ETK K they run a really large offset on the wheel hubs and rims, however, I am not sure why. Must be a good reason, I have never done the testing myself though.
     
    • Like Like x 1
  7. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Hmm, for me it looks like hub nodes are 0.225301 apart each other, while hub width and tire width are 0.225000, is there some other place for that than definition in front wheels ?

    ["axf3", 0.910143, -1.35254, 0.765663{ "group":"axf_L"}],
    ["axf4", 0.684842, -1.35254, 0.76173{ "group":"axf_L"}],

    {"radius":0.350000}, //Radius of the tire nodes
    {"hubRadius":0.266700}, //Radius of the rim
    {"wheelOffset":0}, //Offset from the original axle position (Left/right)
    {"hubWidth":0.255000}, //Width of the rim
    {"tireWidth":0.255000}, //Width of the tire
    {"numRays":16}, //The amount of nodes to make the
     
  8. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    The hub and tyre are 0.255 wide, not 225. Its difficult to show on a still image, but you don't need to look at numbers to see the centre of the wheels is narrower.
    upload_2018-7-29_0-33-37.png
    If you look at the vehicles front left wheel (on right side of image), you should be able to see.
    --- Post updated ---
    Just gave it a test with the following:
    Code:
            ["axf1", -0.66, -1.35254, 0.76173{ "group":"axf_R"}],
            ["axf3", 0.93, -1.35254, 0.765663{ "group":"axf_L"}],
            ["axf4", 0.66, -1.35254, 0.76173{ "group":"axf_L"}],
            ["axf5", -0.93, -1.35254, 0.765663{ "group":"axf_R"}],
    Seems to reduce wobble from an admittedly unreasonably short visual test. Whether it will fix your FFB woes, not sure.
     
    #8 aljowen, Jul 29, 2018
    Last edited: Jul 29, 2018
    • Like Like x 1
  9. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I blame tired eyes :D

    I did a test and sadly there still is hammers banging on wheel, but clearly it is lot less with your values. Also noticed hammering on braking, so it has something to do with hubs being too loose, thx!

    Simple solution would of course be to make a note telling car is not allowed to use on pavement, only on gravel :)

    Oh, need also edit tread coef of tires as currently grip on gravel is not decent.

    Maybe testing different hub widths could make further improvements, or perhaps ffb gain settings to make forces bit less as they are bit strong to my liking, more testing needed.
     
    • Like Like x 1
  10. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    If you need to mitigate it further, you could try increasing the BeamSpring values of the wheel hubs. If this causes the vehicle to become unstable or vibrate a little, add some more weight on the wheel hub nodes (all nodes hanging off the end of the wishbones essentially).

    Of course you may want to avoid adding to much weight onto the hubs, since they are an active suspension component, so the ride quality will become less supple as they get heavier. This may or may not be realistic depending on your car.

    Increased BeamSpring will make the vibrations less apparent since it makes the hub stiffer. However, in BeamNG you can only make beams so stiff before they explode, adding weight onto the nodes mitigates that effect.
    --- Post updated ---
    Small question, does having the FFB debug app open while trying to view the parts or tuning menu for the ETK K cause your games UI to completely lag out and drop down to less than 1 frame per 10 seconds?
     
    • Like Like x 1
  11. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I need to try that next, thx!

    I put these in and it was slightly better still.
    ["axf3", 0.9225, -1.35254, 0.765663{ "group":"axf_L"}],
    ["axf4", 0.6525, -1.35254, 0.76173{ "group":"axf_L"}],

    Then I lowered FFB effect "FFBcoef":"$=13*0.700000", //"FFBcoef":"$=13*0.900000"

    Which made hammering at least less annoying, can't lower that more as FFB is then too light and it is really just masking the issue.

    So need to play with hub stiffness next.

    Also one thing that came to my mind is that if steering rack shakes, would that mess with FFB as FFB is calculated from nodes in hub? I haven't checked if the rack shakes though.

    Increasing beam spring on axle seems to make hammers worse and decreasing seems to make it better, interesting.

    About K-series and UI lag, yes, it does make UI lag quite bad indeed, I guess you found a bug right there :)
     
  12. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    I'm not sure. I don't know whether it cares what the steering hydro is doing. Since the only nodes specified are the two node at the end of each wishbone where the hubs attach. But perhaps the game automatically knows the steering is there and takes it into account.

    When I say stiffen up the hub Beams, I mean all of the hub beams as shown in the image below, not just the other beam that afaik is also known as a wheel hub in beamNG :p
    upload_2018-7-29_1-19-27.png

    More strength in that octahedron should reduce the amount that the wheel can wobble side to side. Essentially making it more ridged, and more solidly connected to the wishbones.

    So essentially:
    Code:
    //double wishbone suspension
            {"beamPrecompression":1, "beamType":"|NORMAL"},
            {"beamSpring":8811956.000000,"beamDamp":260.363831}, <=============== That beamSpring there (the new value I have here seems to work without any node weight changes)
            {"beamDeform":63528.769531,"beamStrength":365550.812500},
            //hub
            ["fh1r","fh2r"],
            ["fh1l","fh2l"],
            ["fh1r","fh3r"],
            ["fh1l","fh3l"],
            ["fh2r","fh3r"],
            ["fh2l","fh3l"],
            ["fh1r","fh5r"],
            ["fh1l","fh5l"],
            ["fh2r","fh5r"],
            ["fh2l","fh5l"],
            ["fh3r","fh5r"],
            ["fh3l","fh5l"],
     
    • Like Like x 1
  13. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Yeah, I did play with those too, while stiffening of axle beam makes it worse, stiffening of other hub beams makes it better, however increasing hub node weight makes it worse, so there is only so much room to improve until I need to increase node weight.

    Beam damping seems to help too, so maybe trying to find maximum for that would work.

    Then again, even hopper tends to hammer the wheel at times, but on flat level ground it is less and effect is less strong at similar FFB strength.
     
  14. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    I haven't downloaded the zip (yet), but I thought I'd throw this out there:

    I think that during his development of the DH Outlaw @Darren9 determined that he just couldn't get a good result as long as he had big front wheels attached to the drivetrain. I don't recall for certain, but I assume that he consulted with the devs on it.

    "ForceFeedback unfortunately can't work well with wide high powered front wheels, try the rear wheel drive configs and/or adjust the ForceFeedback slider in the tuning section to a tolerable level." via https://beamng.com/resources/dh-outlaw.2673/

    Could that be the issue here?
     
    • Informative Informative x 1
  15. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    It is RWD mid transverse engine, not sure if front wheels are that huge really, just 255 wide sport tires/wheels, I would think that even Hopper has bigger and as engine makes only around 330hp and torque is modest (3.5l high revving V10), I'm afraid that is not the cause.

    I'm quite certain that it is something AM exporter has done, it is AM mod vehicle, so it might be that exporter is not getting quite perfect result with it.

    Attached vehicle is probably quite many iterations behind of current one, but that might be just good, not so sure my edits have done as much good as they have done harm, at least they have made new issues, so probably going to revert to attached version.
     
  16. torsion

    torsion
    Expand Collapse

    Joined:
    May 31, 2015
    Messages:
    1,600
    Gotcha. Darren9's problem is definitely only with AWD/FWD/4WD, not RWD. Sorry for the distraction! :p
     
    • Like Like x 1
  17. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    It kinda feels like if car just wants to be hammering steering wheel. Some changes are changing it a bit, but still it is hammering more than other vehicles.

    I don't know how you feel it with a wheel, is it too bad or is it acceptable hammering? Ok, why do I even ask, should be able to fix these things, but pulling my hair with this.
     
  18. aljowen

    aljowen
    Expand Collapse

    Joined:
    Oct 21, 2012
    Messages:
    1,677
    So, I have got the wheel plugged in and tested with FFB.

    Immediate thing I checked was what the FFB coef was set to. "13*0.9" seems way too high. The ETK K uses 2.5, my mods are all below 5. So that set off immediate alarm bells for me. So I changed it to be 5, since that is closer to what I would expect. I am pretty sure the hammering you are noticing is caused by having that too high.

    So the next thing I thought is "what is causing the FFB to be so numb that it can't be felt". So the first thing I checked was the wishbones, since they are what the FFB nodes are attached to.

    The beam values that they were set to were very odd. Completely unlike any other vehicle I have seen in game to be precise (this is first Automation vehicle I have looked at however). I have a mod that I am working on that uses double wishbone suspension in the front, so I made the values on your vehicle closer to mine, this seems to have helped.
    Code:
            //upper wishbone
            {"beamSpring":10000000.000000,"beamDamp":150},
    This changes both upper and lower wishbones to more sane values.

    I also changed the front hub values to make them a little more in line with what I would expect, to see if it helped with anything.
    Code:
            {"beamSpring":9000000,"beamDamp":150},
            {"beamDeform":63528.769531,"beamStrength":365550.812500},
            //hub
    I would probably also recommend going through some of the rear suspension beams and checking/updating them, you may notice the vehicles handling improving due to having values closer to the official vehicles.
     
    • Informative Informative x 2
  19. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Some days are fine, some other days my brain just does not compute, but this reply of yours I think has kicked my brain to realize something I can't understand why I did not realize it, once again, so big thanks to you again :D

    I put some colors in, pink/purple is secondary beams, those were set really soft, something like 3 million, white are subframe, yellow are primary beams. Secondary and primary are subframe mounting to body.
    upload_2018-7-31_8-31-48.png
    Blue ones are A-arms, red ones are tierods and thick green ones are steering limiters.

    So as I now finally realized that comparing to official vehicle is what I should do (stupid brain), then I opened ETK800 and examine values a bit:
    -hub and attach to wheel is 7.8 million (few attach to wheel beams are 12 million)
    -Half of lower arm is 20 million other half is 15 million, there is no upper arm as it is strut.
    -firewall is 2.5 million

    SBR4 then:
    -Hub etc 9 million
    -a-arms 12 million
    -subframe where arms connect to, 6 million
    -firewall is 2.5 million

    Mine has carbon fiber construction, so it should be bit stiffer than steel construction used in SBR4 I guess, but I did set SBR4 values.

    Then I explored bit more, steering and especially FFB:
    Steering arm in SBR 4 is 18cm long, that with 12 FFBcoef, while AM export has 13*0.9=11.7 FFBcoef with 14cm steering arm, if I remember steering arm length correctly as I did shorten it and lowered FFBcoef to have bit faster steering.
    My perfectly scientific method has been now to drive this AM export, then K-series, adjust FFBcoef, drive again until it is somewhere close there, even though it should be possible to calculate, but tires do make difference of course.

    Not very huge difference over my previous adjustments.

    Next I copied steering damper settings from SBR4 and drove SBR4 and my AM export one after another, there still is more hammering with my export, there was clearly improvement, but still slight hammering happens.

    Comparing wheel alignment, looks like I would like to have bit more camber, higher Caster is of course giving more feedback and making steering bit stronger to countersteer by FFB:
    upload_2018-7-31_10-1-37.png

    Need to check and compare more of those values, maybe I missed something, or maybe need to mess with wheels a bit, but somehow this car just likes to hammer the wheel :p

    Oh but it is wonderful that these exports have those coloring codes pre-set as well as lot of useful comments in their files, there is wealth of information and doing these really helps to understand jbeam much better.
     
  20. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Ok, I made another car, different body, different engine, but similar style and it is even worse in this hammering thing.

    I did test out all kind of front suspension types, power steering types, nothing changes it.

    So probably this is something that needs to be fixed in exporter itself.

    I try to test more of first vehicle later though, but just a heads up that there is something in such shaped bodies that exporter can't quite handle.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice