1. Trouble with the game?
    Try the troubleshooter!

    Dismiss Notice
  2. Issues with the game?
    Check the Known Issues list before reporting!

    Dismiss Notice
  3. Before reporting issues or bugs, please check the up-to-date Bug Reporting Thread for the current version.
    0.30 Bug Reporting thread
    Solutions and more information may already be available.

Guide to FFB

Discussion in 'Troubleshooting: Bugs, Questions and Support' started by fufsgfen, Aug 2, 2018.

  1. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    First one needs to read and understand this (well, I'm not sure if even I understand it completely):
    https://wiki.beamng.com/Steering_wheel_setup

    For T150 wheel, I had to set FFB update type to Full or there was no countersteering, check that first, use K-series for all this to get a baseline setting, which you can then change to other vehicles if needed.

    There are some challenges in setting FFB, so often I see people maxing out FFB level, then complaining that they can't feel grip loss etc. There is natural explanation to that, FFB clipping, which occurs when you are asking wheel more force than it can produce.

    There is certain point, which after all the forces feel same and it comes surprisingly early, feeling is very misleading quite often.

    You will need to add FFB app to see graph in below picture (it makes part selector really laggy, so when you have graph on don't open parts selector).

    Open this image to full size so you can read text in it:
    upload_2018-8-2_7-19-31.png

    For example if your wheel is maxing out at point 1, point 2 will feel same as point 1, so when force decreases from point 2 to point 1 there is no change that can be felt either.

    Sadly wheel are not perfect, so for example strong countersteering when drifting or feeling smallest of bumps, might require you to set force quite strong, which means it clips quite early and thus you can't feel small changes in grip during cornering.

    On the other hand, you might get really nice feeling of loosing grip during cornering, but you don't feel any of smaller bumps, also countersteering might feel weak and slow.
    Wheels tend to have certain minimum force which is needed until they produce anything that can be felt, I believe they call it FFB deadzone.

    So setting wheel is compromise and you might need to set up it differently depending what kind of driving you are going to do, that is case with my cheap T150 Thrustmaster, better wheels are better and can do greater ranger of things right without re-adjusting, surprisingly.

    upload_2018-8-2_7-34-6.png

    Following methods are how I think one could achieve good settings, but I know only T150 wheel and only about two weeks, so feel free to share your thoughts.

    1. First set Strength to quite strong, maybe 400 or so, filtering of both at least half way, point here is to set wheel so strong as it will go.

    2. Next set Maximum strength to something what you think your wheel might be, G29/G290 2.1-2.2 should be somewhere there according to http://edracing.com/edr/Wheel-Torque.php
    NOTE: it looks like that maximum strength is not Nm, despite FFB App lets one believe so, currently not more is known about it, but try to ignore Nm in FFB app.
    It could be that maximum strength slider is percentage of what your wheel's driver can make happen, so 10 would be 100%, but that is just a guess, more data is needed about it.
    Something between 2 and 3 might be good place to start experimenting with instructions given below.

    3. Now you go driving, test all most common cornering situations, spend good time driving so that your hands will get used to really stiff steering.

    4. Reduce Maximum Strength by 0.5, go driving again, is there difference? If you can feel strength being less, change up by 0.2 and test again, if there was no change, change down another 0.5.
    Idea here is to find out what is maximum strength your wheel can do, correct setting is right there where increase does not change anything and where decreasing changes the strength.

    Next different setting methods are for wheels like G29, T150 etc. those with good strong wheels, probably would be okay to set wheel for driving and be done with it.

    5(Racing). After finding your maximum strength, your goal is to have clipping so small level that you can feel grip loss, so that during normal cornering you will not have force quite maxed out, but have little bit of extra there for wheel fade and for different situations where you might experience bit harder.
    So you keep lowering Strength setting until you are seeing something like 80-90% of maximum Strength during normal cornering.
    (Note: FFB probably is lot weaker than you are used to, use it quite long time to get used to it as when you get used to it, you should find it better than strong one)

    5(Drifting). After finding your maximum strength, your goal is to have clipping at tolerable level, while still maintaining some feeling of grip loss, probably around 100% of max strength is needed for common wheels, that it is at least for my wheel.

    It can be that optimal setting is somewhere between the two.

    Here I'm testing if I can feel grip loss when turning too much, make sure you get 60fps when doing this as lower FPS can mess with FFB.


    FFB Smoothing
    There are two smoothing sliders, fast and slow, when you turn your wheel quickly fast, that is what fast smoothing affects and slow turning or going straight for slow smoothing.
    If wheel feels grainy/coarse/ like it would have gears with big notches in them, more smoothing is needed, but too much can be bad.
    Also if you go really low on smoothing, it does make wheel weaker, at least that is how T150 seems to behave, so you might need to adjust your strength setting when lowering this a lot.

    Strength at slow speed is to deal with violent shaking when moving slowly, start with same setting as Strength, then lower if there is such violent shaking, until shaking is no more.

    Use response correction curve
    This makes my wheel to have very very narrow resolution, I can't feel much of any difference in FFB, despite I did LUT curve according to instructions for AC which I guess should work, puzzled about this one.

    Update rate
    From some reason when Drifting with 200BX, I get somewhat more fine details with 2000Hz, which should certainly not be setting for cheap wheel like T150, but it is not always good at 2000Hz, Automatic or even lowering to 100Hz works sometimes better, depending car, driving etc.
    Yeah, I'm puzzled about this one too, maybe we learn more about those someday.


    Thoughts, comments, correction, word is free, I have only about 2 weeks with a wheel in BeamNG, even I did wear out two MS wheels, DFP and G25 in history, T150 is only one I have used with BeamNG and Beam is bit different from racing games, so there could be things to improve, but I hope that this helps some to have more fun out of their game and equipment.
     
    #1 fufsgfen, Aug 2, 2018
    Last edited: Aug 11, 2018
    • Like Like x 5
  2. rottenfitzy

    rottenfitzy
    Expand Collapse

    Joined:
    Mar 7, 2015
    Messages:
    680
    Is there any way to make the wheel heavier overall? My g920 feels awfully soft to turn. Is there any way to fix that?
     
  3. Nadeox1

    Nadeox1
    Expand Collapse
    Spinning Cube
    BeamNG Team

    Joined:
    Aug 5, 2012
    Messages:
    14,683
    Tried to increase the strenght?

    Also, nice guide! Added here
     
    • Like Like x 2
  4. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    I would guess Max strength 2.1 or 2.2, check from link in OP which it is for g920.

    Then more strength as Nadeox mentioned, you might get some clipping, so don't overdo it, it is bit of compromise to get good sharp force and avoid loosing higher strength ffb details. Pay attention to FFB graph when driving so you don't get too much clipping.

    However for drifting I do like to set force bit too strong so that I loose details on stronger forces, but as a tradeoff I get sharper countersteer. Fanatec would be so nice, but we have to deal with what we have :p

    Also if your filtering is very low, FFB tends to get soft and weak, at least with T150, maybe wheel misses some fast changing data, so increasing filtering might be worth to try, slow for steady cornering and fast for quick movements of the wheel.

    Assetto Corsa has minimum force, I'm not sure if it actually does lift lower end forces while keeping higher end forces, kinda truncating all forces to smaller range which starts up higher, that might help dealing with FFB deadzone, however I'm not all too certain about all technicalities involved, BeamNG devs might have better ideas than such minimum force in cooking, but we see when they push new features in updates to FFB which I did read is still area of WIP.
     
  5. stenyak

    stenyak
    Expand Collapse
    Programmer & Global Moderator
    BeamNG Team

    Joined:
    Aug 6, 2012
    Messages:
    1,891
    Some explanation about this. It's simplified but should give you an overall idea.

    Due to various reasons, the sim sends FFB requests at a variable rate. Even if you set 2KHz, it won't always be a perfectly smooth constant stream of 2KHz updates. There might be bursts of e.g. 10Khz, and then periods of e.g. 0.5KHz.

    Now, not all steering wheel models' drivers/firmware/hardware is able to cope with any update rate. Some will choke with a burst of 10KHz update. Some won't be able to handle even a 0.1KHz burst (true story).
    Possible side effects: you get random, full or zero forces, you get lower framerates, other features on the steering wheel stop working, or work incorrectly, etc. I've seen all kinds of things, and some were very subtle.

    The "Update Rate" setting makes sure that, no matter what, we don't overload the steering wheel. If you choose 2KHz, then the sim will never, ever, use an instant rate higher than 2KHz. Sometimes it will be precisely 2KHz, sometimes it will be lower than that. It's a safety cap, a maximum limit.

    The "Automatic" update rate uses a mechanism to determine a probable safe rate for your current combination of hardware and software. It works with most of the devices out there, but it's not 100% bullet proof. The "Update Rate" setting was therefore added to our menus, so people with this 1% of problematic devices can experiment and manually lower the rates when needed.
     
    • Informative Informative x 2
    • Like Like x 1
  6. B. Tanner

    B. Tanner
    Expand Collapse

    Joined:
    Jan 20, 2015
    Messages:
    92
    Do the numbers of the Maximum strength setting really represent torque in Nm? It was my first guess, too. But the longer i fiddle around i am not too sure anymore.
    Its just a bit confusing. With the G27 i can get away with a "max strength" of 2.1 and "strength" at ~600. But also with a "max strength" of ~3 and "strength" set to ~200. Both settings provide similar torque but the information the wheel provides feels quite different. Clipping is okay with both. Here and there some spikes but nothing you cant finetune with smoothing.

    Either way limiting the max. strength seems to be a good idea on lower end wheels even tough the wikipage states the exact opposite. The effects of this setting probably changed with the last update. Especially on hard landings and crashes the g27 cracks and rattles violently when max. strength is set on 10.

    /Oh, totally forgot: Thanks for the guide. Its a great write up.
     
    #6 B. Tanner, Aug 10, 2018
    Last edited: Aug 10, 2018
  7. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Thx!

    Max torque is Nm, if you enable FFB app, you can see it adjusts two dark red lines in chart which are Nm and correspond to value you have max torque set at.

    I have strength at around 100 now, bit less or bit more depending what kind of driving I do and which vehicle.

    Wheel check gave me 2846 as max force wheel can produce, but I have no idea of what Nm value that would be. I know when I set game to use correction curve, that max torque pink graph will show is something like 1.6-1.8, yet I can definitely tell difference of 2.2 and 2.4, 2.6 is maybe.

    I test that by setting node grabbers to pull wheels and crank force quite high, so that I will get force that is at limit, then I move wheel with different max strength settings and try to find what is maximum where I can feel difference.

    So indeed, there are some question marks over that subject. While max strength is in NM, is Nm shown same as Nm that are quoted being what for example G29 can produce?

    I think that I have read those quoted torques of wheels being something around there more than accurate values, so better test yourself what is maximum where max force changes force amount.

    Now without testing, if you just set max force to something and then crank up force, it is very easy to get mislead, because humans are quite subjective beings. What might feel stronger, might actually be same, but just ramps up earlier, which is why I set max force as described.

    After setting max force, I can then find lowest force strength that will produce that max force under hard cornering, to get some baseline where to tune with subtle changes.

    I have seen people using very high force and smoothing, but that results mostly digital FFB, sure it feels strong and when drifting is quick to turn, but feeling all the subtle changes in grip, like understeering is lost then, that is then ones own choice if one wants to make such compromise.
     
  8. Goosah

    Goosah
    Expand Collapse
    Global Moderator
    BeamNG Team

    Joined:
    Aug 30, 2013
    Messages:
    790
    Are you sure about the max strength value? I tested my g27 by setting the force cap to 10 and turning the wheel at rest while increasing gravity, I found that the force in the g27 was not physically clipping until the forces were reaching near the force cap. I don't think the cap is actual Nm at all, its whatever the driver decides it is.
     
  9. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    All I know is that it adjusts these dark red lines and it does say Nm in there and maxStrength number corresponds to number shown on those limit lines:
    upload_2018-8-11_5-48-35.png

    How well that then translates to wheel torque, what driver puts out to wheel, I have no idea.

    Trouble is that wheel can feel stronger if forces ramp up faster, clipping is something that is very hard to tell apart under normal driving because one might feel wheel being stronger while in reality it just ramps up force faster to maximum.

    Also I can tell that if I set maxStrength to maximum, it is bit stronger than 2.6.

    So I can't be sure of anything really, it says Nm there next to Limit yet it kinda is not Nm, so maybe we just need to ignore Nm on graph and just rely on testing so we can get wheels set so that they don't clip forces when driving, so we can get maximum force and resolution to get maximum amount of data from the FFB?

    Would maxStrength as well as strength settings be described as a percentage of what driver will do? When adjusting them around, it surely feels like if they would be percentages, even maxStrength and on FFB graph everything is just percentage of maximum possible?

    Then one question begs asking, why there is Nm in graph if it is not Nm? :D

    I did sent lot of stupid questions to Stenyak yesterday, and now I learned again bit more, maybe we all learn something of this :D

    If maxStrength just is percentage of maximum driver will send to wheel and FFB graph is just percentage of maximum setting of maxStrength, then it would make sense to have all settings as percentages?
     
    #9 fufsgfen, Aug 11, 2018
    Last edited: Aug 11, 2018
  10. stenyak

    stenyak
    Expand Collapse
    Programmer & Global Moderator
    BeamNG Team

    Joined:
    Aug 6, 2012
    Messages:
    1,891
    Some clarifications.
    The use of Nm in the app is indeed wrong. In theory, we might be able to use plain Newtons (as measured at the contact point between hand and device during normal operation of the wheel/joystick/etc). In practice, all steering wheel drivers I have been able to test, lie with regards to these units. Let alone if you use custom rim diameters, etc. So all bets are off right from the start. I will remove the label to prevent confusion.

    Regarding the "Max strength" value. This is supposed to be measured in Newtons, but after testing numerous wheels, all drivers seem to share a value of "10" as their maximum force capacity. You can verify this in the game logs, look for "ff_max_force". Please do let me know if anyone ever sees a value different than 10 in the logs. We simply don't have the resources to test all steering wheels in the market.

    In addition, there's the parameter of "ff_res" (right below, in the logs). This is the granularity of force provided by the drivers. A typical value is 256, which means 256 different possible force values you can request (within that range of 0 to 10). We have decided to trust this information, but we don't have a dynamometer to verify its correctness.

    So, which of those 256 (or however many) steps provide a measurable difference, is a different story. You can generate a response correction curve to get a rough idea. These tools are not perfect, and additionally they have to work with very limited information. But you should see if your wheel response is flat in the upper range, if there's a dip in the middle, or a deadzone at the beginning. I repeat, this is not fully trustworthy information, but many times, it's better than nothing.

    Enabling the use of this curve in BeamNG.drive should generally correct the situation and provide a fully usable 0-to-10 range. But as you see, this depends on so many assumptions and on so many factors, that it may not be the case always.
     
    • Informative Informative x 1
  11. fufsgfen

    fufsgfen
    Expand Collapse

    Joined:
    Jan 10, 2017
    Messages:
    6,782
    Yeah, more I test these things, more I find out that what works on Logitech, clearly needs something else with this T150. So I think testing and analyzing this one completely might provide some useful data.

    That information about log values is very useful, mine looks to be like them are typically.

    Also I found some claims on internet that on Thrustmaster 70% at windows driver settings is what provides maximum torque, which is quite different from Logitech, however that was regarding higher end Thrustmaster. That would be so that setting higher than 70% only makes lower forces stronger, so all forces are kind of truncated so there is less available difference between low and high.

    I will try to find some nice low cost fish weighting equipment, which should work in measuring resistance of wheel in kg, also it should give some indication of different force settings and help finding out at which setting strength is at maximum, but of course can be used only vehicle stationary.

    T150 256 steps, 0-10 000:
    upload_2018-8-11_16-34-23.png
    Maybe FFB just is completely junk in this? Then again without correction curve it does not feel hugely different from G25 I had.


    If I set FFB max strength to 5 instead of 10, do I get 256 steps between 0-5 or 128 steps? If I set Windows driver to 70% I guess that would be 7 of maxStrength, would it be with 256 steps or will it always be 256 steps to 10 000 and reducing strength or setting max strength just reduces number of steps too? I guess each step is locked to force level and less max force / force strength just does not use higher steps, but tbh I have no idea.

    If these consumer devices would just have spec sheets like industrial devices, our lives would be so much easier :p

    I think that this is very valuable discussion as it increases understanding of FFB and benefits whole community, thanks from your patience! explaining these aspects :)
     
  12. stenyak

    stenyak
    Expand Collapse
    Programmer & Global Moderator
    BeamNG Team

    Joined:
    Aug 6, 2012
    Messages:
    1,891
    This is why I say the correction curves are not necessarily valid. My suspicion (which would require writing my own wheel tester tool verify) is that you need more than this simplistic correction curve, you need a correction surface, by repeating the measurement with 5, 10, 20 different starting rotational speeds on the wheel. These testers all start from a stationary position, which (as I mentioned in our private conversation) is probably biasing the measurements incorrectly, ignoring the initial static friction. But again, it's just a suspicion. It is possible that your specific steering wheel model has a more pronounced effect there (because of its design, material properties, etc), making the correction curve useless, or needing huge tweaks before it can improve the situation. It might also be the case that the wheel/drivers are already tuned from factory to work better in dynamic conditions, even at the expense of worse static behaviour (as is reflected by these tools). Who knows.


    You get half the steps, assuming your response curve is fully linear or disabled (same thing). With a non-linear curve, you might not get 128 steps, but maybe 150, or 100, etc, depending on the actual shape.
     
    • Informative Informative x 1
  13. Trophy

    Trophy
    Expand Collapse

    Joined:
    Aug 26, 2019
    Messages:
    1,644
    I cant find fast and slow smoothing. All there is is smoothing
     
  14. esesel

    esesel
    Expand Collapse

    Joined:
    Feb 18, 2019
    Messages:
    741
    This is a old thread it has changed by now
     
  15. stenyak

    stenyak
    Expand Collapse
    Programmer & Global Moderator
    BeamNG Team

    Joined:
    Aug 6, 2012
    Messages:
    1,891
    • Like Like x 1
  16. Trophy

    Trophy
    Expand Collapse

    Joined:
    Aug 26, 2019
    Messages:
    1,644
    deleted
     
    #16 Trophy, Jun 26, 2020
    Last edited: Jun 26, 2020
  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