Just figuring out various details about rF2 geometry

Discussion in 'Car Modding' started by mantasisg, Oct 8, 2020.

  1. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,926
    Likes Received:
    3,876
    (#1) Is there a way to imitate or simulate De Dion tube in rF2 ? One question that follows from that - (#2) is rF2 suspension geometry modular, or there are only some strict combinations of linkages that can be recognized by physics engine ?

    This is how basic De Dion layout looks like:


    (#3) Could I just use a bar and link it from one spindle to other spindle ?

    This is how standard tracking bar parameters look like:

    [Constraint]
    Type=Bar_2
    Name="TRACKING_BAR"
    BodyA="REAR_SUBBODY"
    BodyB="REAR_AXLE"
    SpringStability=1
    SpringRate0=100000000
    DamperRate0=100000
    Motor0=0
    MinForceOrTorque0=-9000000000
    MaxForceOrTorque0=9000000000
    AccumImpulse0=0
    MinError0=20
    MaxError0=20
    AutoError0=0
    Contacting=0
    Breakability0=0
    LocationSolving=1
    LocalOffsetA=(0.45999999999999963,-0.08000000000000009,0.49707999999999997)
    LocalOffsetB=(-0.45999999999999996,-0.033000000000000036,0.11200000000000013)

    I wonder could I use it like this:

    [Constraint]
    Type=Bar_2
    Name="TRACKING_BAR"
    BodyA="RL_SPINDLE"
    BodyB="RR_SPINDLE"
    SpringStability=1
    SpringRate0=100000000
    DamperRate0=100000
    Motor0=0
    MinForceOrTorque0=-9000000000
    MaxForceOrTorque0=9000000000
    AccumImpulse0=0
    MinError0=20
    MaxError0=20
    AutoError0=0
    Contacting=0
    Breakability0=0
    LocationSolving=1
    LocalOffsetA=(0.45999999999999963,-0.08000000000000009,0.49707999999999997)
    LocalOffsetB=(-0.45999999999999996,-0.033000000000000036,0.11200000000000013)

    (#4) Bit silly question, can I use any name I like for the constrains, for example "PANHARD_ROD" or really anything ?

    Swing axle questions. (#5) Is there a way to simulate swing axle specificaly, or is it considered enough to do by using standard double wishbones links geometry ? (#6) Could I simply remove upper rear links, or even both upper links ? Additional question, related to suspension modularity question (second qustion): (#7) If imitating swing axle by using double wishbones constraints I would add extra bar like tracking bar, would it work ?

    Some live rear axle questions similar to above questions. Right now I have only used two types of live rear axle in rF2 which I am using at the moment. First - live rear axle of the iconic box master pizza delivery car, which uses: tracking bar, single left and right trailing links and middle control link. Second - live rear axle from ISI_interceptor oddity, which uses lower and upper, right and left trailing control links and tracking bar, (minus swing_arm element, which I suppose could do something similar to dion tube, but I am more than uncertain about that, (#8) could it ?).

    There are few pictures of impressive quality Dtype scale model. It can be seen that there is no tracking bar, but lateral stability job is done by these triangular truss like links. (#9) I wonder could I remove tracking bar from rF2 chassis geometry, and add two links from rear subbody to rear axle and get functional kinematic system ?
    [​IMG]
    [​IMG]

    Well, ifall that is not enough, I also have a question about this thing. This is scheme of Maseratti A6GCS rear suspension with quarter-elliptic leaf spring lower rear trailing link. (#10) I wonder where the correct body pivot point would be for such element, I suppose it would have wrong arc if it would be set at same position in rF2 chassis geometry. Also what would be correct position for PushrodSpindle to PushrodBody link, straight up vertical from center of the spindle ?

    [​IMG]


    Thank you if you bother with this, if you do.
     
  2. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Yes there is. The axle trailing arms are like a normal beam axle (live axle), but subtract the unsprung mass of the differential. The additional aspect of the emulation is a 10 meter panhard rod that is anchored to the middle of the axle (instead of the right hand end of the axle), parallel to the ground, and fixed to the body at the other end (it's a virtual connection). The 10 meter length means that the motion error is <1mm over most of the axle's vertical movement.
     
    Last edited: Oct 8, 2020
    mantasisg likes this.
  3. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Normally an upper and lower wishbone are used to emulate a swing axle because the physics engine needs a pair of wishbones. They share a common point at the differential end and are usually a narrow base. Low pivot point swing axles (e.g. 1955 Mercedes) can be emulated and there's an example in the rF1 '55 F1 mod.
     
    mantasisg likes this.
  4. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    As I recall, the track bar (panhard bar) is optional. It's main reason for existence in the rFactor suspension model is to account for sideways movement of the axle as it moves up/down. The two angled bars of the D-type are probably best represented by a single bar at the bottom.
     
    mantasisg likes this.
  5. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    It's been mentioned that the torsional resistance feature can have multiple sections (rather than just the single section we commonly use), but I have no idea if it would be applicable here and how one can even build multiple sections. I think there's enough slop in leaf springs that placing the pivot point at the edge of the body, the edge of the clamp, is adequate.

    In terms of Pushrods and spring rates, yes. Certainly simplifies the math!
     
    mantasisg likes this.
  6. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Here's a De Dion or Watts Link example from the rF1 Caterham mod:
    [BODY]
    name=rl_wheel mass=(10.0) inertia=(1.292,0.724,0.724)
    pos=(0.668,0,1.1125) ori=(0.0,0.0,0.0)

    [BODY]
    name=rr_wheel mass=(10.0) inertia=(1.292,0.724,0.724)
    pos=(-0.668,0,1.1125) ori=(0.0,0.0,0.0)

    [BODY]
    name=rear_axle mass=(30.0) inertia=(0.0275,9.25,9.25)
    pos=(0, -0.113, 1.1125) ori=(0.0,0.0,0.0)

    [JOINT&HINGE]
    posbody=rl_wheel negbody=rear_axle pos=rl_wheel axis=(-1.00,0.0,0.0)

    [JOINT&HINGE]
    posbody=rr_wheel negbody=rear_axle pos=rr_wheel axis=(1.00,0.0,0.0)

    //Kangaloosh link - body mounting is 10m to left of car, simulates Watts linkage in rFactor
    [BAR]
    name=track_bar posbody=body negbody=rear_axle pos=(10,-0.113,1.1125) neg=(0,-0.113,1.1125)

    //Torque link left
    [BAR]
    posbody=body negbody=rear_axle pos=(0.65,0.037,0.1125) neg=(0.65,0.037,1.1125)

    //Torque link right
    [BAR]
    posbody=body negbody=rear_axle pos=(-0.65,0.037,0.1125) neg=(-0.65,0.037,1.1125)

    //Torque link centre
    [BAR]
    posbody=body negbody=rear_axle pos=(0,-0.263,2.1125) neg=(0,-0.263,1.1125)
     
  7. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Here's the low pivot Mercedes example from the rF1 Team Players 55F1 mod:
    // Rear left suspension (2 A-arms + 1 straight link = 5 links)
    //Note have changed -0.105
    [BAR] // forward upper arm
    posbody=body negbody=rl_spindle pos=(0.0,-0.105,1.004) neg=(0.563,0.113,1.104)

    [BAR] // rearward upper arm
    posbody=body negbody=rl_spindle pos=(0.0,-0.105,1.204) neg=(0.563,0.113,1.104)

    [BAR] // forward lower arm
    posbody=body negbody=rl_spindle pos=(0.0,-0.105,1.004) neg=(0.561,-0.09,1.104)

    [BAR] // rearward lower arm
    posbody=body negbody=rl_spindle pos=(0.0,-0.105,1.204) neg=(0.561,-0.09,1.104)

    [BAR] // straight link
    posbody=body negbody=rl_spindle pos=(0.0,-0.105,1.004) neg=(0.561,-0.09,1.004)


    //----------------------------------------------------------------------------------------

    // Rear right suspension (2 A-arms + 1 straight link = 5 links)
    [BAR] // forward upper arm
    posbody=body negbody=rr_spindle pos=(0.0,-0.105,1.004) neg=(-0.563,0.113,1.104)

    [BAR] // rearward upper arm
    posbody=body negbody=rr_spindle pos=(0.0,-0.105,1.204) neg=(-0.563,0.113,1.104)

    [BAR] // forward lower arm
    posbody=body negbody=rr_spindle pos=(0.0,-0.105,1.004) neg=(-0.561,-0.09,1.104)

    [BAR] // rearward lower arm
    posbody=body negbody=rr_spindle pos=(0.0,-0.105,1.204) neg=(-0.561,-0.09,1.104)

    [BAR] // straight link
    posbody=body negbody=rr_spindle pos=(0.0,-0.105,1.004) neg=(-0.561,-0.09,1.004)

    // Rear spindle and wheel connections
    [JOINT&HINGE]
    posbody=rl_wheel negbody=rl_spindle pos=rl_wheel axis=(-1.0,0.0,0.0)

    [JOINT&HINGE]
    posbody=rr_wheel negbody=rr_spindle pos=rr_wheel axis=(1.0,0.0,0.0)
     
  8. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Best answered by referring to Bristow's "bible" for rF1's PM file that was posted on RaceSimCentral back in the old days.

    upload_2020-10-8_11-10-15.png
     
  9. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
  10. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
  11. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
  12. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    During this section, please reference the physics blog to see what has changed between rF1 and rF2. Maybe some day an ambitious person will rewrite Bristow's "bible", collating all the revisions?
    upload_2020-10-8_11-29-30.png
    upload_2020-10-8_11-33-23.png
     
  13. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
  14. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    Beam front axles are important for vintage vehicles... anybody tried it in rF2?
    upload_2020-10-8_11-37-39.png
     
  15. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,926
    Likes Received:
    3,876
    Absolutely massive response, Emery. Thank you ! A lot of food for thought and further knowledge for sure, and hopefully improved results in simulating cars.

    I am eager to try this. Stupid of me is that I knew that something like that was done in boxmaster, and indeed just checked - 20meters away to the left starting from the center. I did try 3 or perhaps 5 meters, but the handling just sucked badly. Although I got quite a lot improvement later because of stronger or strongest differential settings and softer dampers and stiffer rear roll stabilisation. I know from experience that boxmaster has foolproof handling lol, can't wait to check the tweaks in game now. And also looking forward for next time to attempt simulating swing axles with de dion tube in this way + substracting unsprung weight, I suppose the "axle" weight and inertia of de dion tube should also be rather low.

    I have a feeling that lately have been getting rather easily along with the 300sl. May I ask, what do you mean by low pivot point of '55 Mercedes ? Doesn't look particularly low from this cutaway, thats SLR:
    [​IMG]
    seems pretty much same, those lower arms looks interesting, do you mean those ?
    [​IMG]
    However, earlier 300SL freaks me out even more. There seems to be no longitudinal member, I don't understand how it could be so.
    [​IMG]

    And thanks for those pm lines and especially for all the screenshots of Bristows posts, priceless stuff. One particular thing that helped, is that axles inertia is set up in such way that inertia body appears flatter at the bottom and top, I had it other way arround, I suppose that was my mistake as well.

    Thats pizzamaster
    [​IMG]

    Thats quad
    [​IMG]
     
    Emery likes this.
  16. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
    I've attached the notes that were included with the Team Player's mod. That might help explain the low pivot swing axle.

    The ordinary 300SL Coupe is just a normal swing axle with radius rods. Like you, I can't find a good picture.
     

    Attached Files:

  17. Emery

    Emery Registered

    Joined:
    Oct 24, 2010
    Messages:
    3,035
    Likes Received:
    1,654
  18. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,926
    Likes Received:
    3,876
    Thank you a lot of interesting read. I already found some useful information in these sources, but also probably a lot more new questions. Not only about the low pivot thing (maybe it means low instantanious center ?), but also about the tires they have modeled and data such as "decreasing crossply tires slip angle with added load), it is weird.... but it is off-topic...

    P.S. this is bit of offtopic in my own thread, but I am surprised at amount of compromises they had to take with initial '55 release judging by the text. It shows that even back then it was so much important to comply to likes of others and maintain certain standard of comfort - higher grip, engine durability influenced use of unauthentic limiters, steering degrees to adjust to lower end steering wheels hardware, and some more bits... And that is quite unfortunate. By reading about tires, it was quite obvious that car should have had notable aerodynamic lift, but they only referred to tire speed sensitivity, which also just couldn't not be especially with harder rubber compounds, and it wouldn't be as influential as aerodynamics anyway. I wish all these modders wouldn't have skipped on rF2, but life goes on and I don't imagine PC you had to have to run rF2 when it was fresh.
     
  19. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    2,926
    Likes Received:
    3,876
    I was having some troubles with simulation de dion and I think it is good to make a note here. Basically was trying to induce proper amount of roll like in pictures of a racecar and couldn't do it. I was able to make front end roll by lowering front inner geometry points, running very soft springs, dampers and antiroll bars, increasign frotn end grip and increasing tire stifness for tire lateral roll. And at this point my inner rear wheel started coming up in the air, and not by little. I was looking for reason to that eveywhere. Spent some hours on it. Then realised that rear axle needed to simulate de dion was too stiff. But rear axle doesn't have deformation parameters. Altering trackbar stiffness didn't seem to do much obvious results. It took some time to think about the joints, of swing axle after I attempted to induce more torsional chassis flex for subbodies joint and it also didn't seem to make much desired effects. Finally after finding joints connecting rear axle to wheels and lowering its spring and damping effects way lower from 1e999 I finally got desired results. Awesome !

    @Emery I have been doing some more swign axle stuff, and I think I learned in practice what high pivot center for swing axles is :D

    I seems this happens if it is high:
    [​IMG]

    Some other things that I have come to, that maybe some guy is missing out still, like I did:
    • ptool instructions: https://www.studio-397.com/modding-...n-to-physics-tool-ptool-and-flexible-chassis/, which is overall great quick run through ptool with many great key tips on how to work on it and with chassis.ini in general.
    • unsprung masses and inertias in chassis.ini - wheels, spindles, rear axles are making great impact for car handling and overall physics functioning). Which I didn't realise how much untill I got them really wrong for some car. I also now may have a clue why I couldn't make smaller size wheels (physics were crashing or working badly) generated with redapg tools: http://meetme.bplaced.net/rF2_onlineTools/, now will probably be able to use for most sizes. Well maybe...
    • corrected inner suspension heights parameter importance, just search what great knowledge guys have said about it by typic "CISH" in search...
    • Don't forget the tires when working on geometry, they cast forces on the chassis. For example if car doesn't roll it might be because there is not enough bite and it slips instead. Also if car is oscilating, it might be related to tires generating grip at greater frequency than chassis is rolling, in could happen in a range from simple "pilot induced oscilations" to undrivable for anybody. This besides aerodynamic setting is the reason why high grip modern race cars are all stiff like a rock. Classics are the best !

    I have one question about wheel inertias in chassis.ini. It does seem like solver of devmode doesn't put out any changes for wheels and spindles inertias and masses. I suppose it needs to be all manually. Perhaps it is something that ttool can put out ? Or it has to be manually calculated, which I think is rather difficult task. It seems to me like in rF2 it is getting more sensitive and easier to cause errors as the objects gets lighter and smaller. Which makes sense that such properties require more precision.

    Also one question about center of gravity. It is said to be taken from reference plane. Reference plane is always defined by rideheights, yes ? Or is there more to that ? CISH is always *CISH = tire Radius - ride height* if wheelcenters are on the bottom of the grid in ptool, or in other words they positioned at 0 vertical coordinate ? And it does not affect reference plane, or does it ? I have seen some things about it written in forum by searching CISH, but didn't totally understand.
     
    Emery likes this.
  20. Lazza

    Lazza Registered

    Joined:
    Oct 5, 2010
    Messages:
    12,345
    Likes Received:
    6,572
    Reference plane should be 0 on the Y axis, as far as I'm aware.
     

Share This Page