Vehicle dynamics test rig

Wheel support

BeamNG supports PC-compatible steering wheels and has force feedback. The publisher lists 112 tested devices across 30 brands, with USB vendor and product ids, and records two separate facts about each one: whether it arrives with in-game controls already assigned, and whether it supports force feedback.

Those are not the same fact, and the difference is the most useful thing in the table. A device can be fully bound and have no force feedback — that is every gamepad. A device can have force feedback and no bindings at all — that is most direct-drive bases, which is why the publisher’s own troubleshooting page has an entry for a direct-drive wheel that produces no forces. It was never configured, because nothing configured it.

Absence from the list is not a verdict either. An unlisted wheel usually works once its controls are assigned by hand; what the list tells you is whether that work has been done for you.

The hardware chain

Six links carry your steering from the rim to the tyre and the force back again, and every published fault for this game lands on exactly one of them. Three are outside the game entirely.

That is the reason to draw it. A reader whose wheel is being hidden from the operating system by Steam can spend a week in the force feedback menu without improving anything, because the menu is downstream of the problem. The diagram on this page names what each link is and what goes wrong there, so the question becomes which link rather than which slider.

Rotation angle

One figure lives in two places: the manufacturer tool and the game. The publisher’s instruction is to set the largest rotation the hardware offers, then set the same number in BeamNG.

When steering feels impossibly fast — the car unable to be held straight, a small movement asking for a large angle — it is almost always these two disagreeing. It is not a sensitivity setting and there is no sensitivity slider that repairs it. On a G25 or G27 the range can be changed during play, with the two middle red buttons on the H-shifter together with the bottom black button; on other wheels it means closing the game and using the driver options.

Pedals

Take separate throttle and brake axes wherever the hardware offers the choice. The publisher says so and the binding you use depends on it.

A combined-axis wheel uses the single throttle and brake binding, with nothing under throttle or brake separately. A separate-axis wheel uses throttle and brake, with nothing under the combined one. Having both present is its own documented fault: the pedals cannot be set up and the car does strange things.

The bench on this page settles which you have without reading a manual. Assign throttle, assign brake, and if both land on the same axis number they are combined.

Force feedback, in order

The publisher’s setup page is a sequence, not a settings list, and the sequence is the whole of it. Smoothing goes to maximum first — it will feel vague and slow, and that is intended — because detail hides the overall force level and removing the detail is what makes strength judgeable. Only then is strength tuned, and only after that does smoothing come back down.

Tune while driving the way you actually drive. Ignore what the wheel does while parked: parked oscillation is what the final step removes, and judging strength by it tunes for the one case that already has its own fix.

If crashes feel violent, lower the torque ceiling in the wheel’s own configuration tool and raise in-game strength to compensate. That moves the peak a crash can reach while leaving everything below it where you tuned it. Lowering in-game strength instead would flatten the entire range to fix the loudest moment in it.

Frame rate belongs in this list even though it is not in that menu. Force feedback is computed per frame; the publisher recommends 60fps as a floor and says higher is more responsive, worth sacrificing graphics and traffic for.

Response correction curves

This is the one part of the subject that is a measurement rather than a preference.

Ask a wheel for one unit of torque and it may return 1.2. Ask for two and it may return 1.9. The relationship is not linear and cannot be known without testing every value, so a response curve records what your hardware actually does and lets the game request the wrong number to get the right result. WheelCheck and Force Curve Modifier produce one; it goes in settings/inputmaps/ and Ctrl + L reloads it without restarting.

The extension matters. Numbers inside a .lut file mean something different from the other formats, so renaming it is not a conversion — the extension is how the game knows which way to read them.

The game sanitises what you give it: normalised from zero to the wheel’s maximum, valleys removed so the response only ever increases, and both ends pinned. And there is one place where the honest measurement produces the wrong result. Most wheels cannot deliver small torques at all, so the measured curve is flat at the bottom and the game compensates by asking for a great deal — which is what produces low-torque vibration. Adding a slight initial slope by hand prevents it, even though the measurement said the dead zone was real.

A curve belongs to a driver configuration rather than to a wheel. Change the effects strength in the Windows driver panel and it is no longer the same curve. Between two owners of the same model, it usually transfers.

Faults

Fourteen faults are published for wheels and six of them are not a setting: Steam hiding the device, a driver window left open, a console mode on the hardware, a binding added to the wrong one of two similarly named entries, a feature that was removed deliberately.

That ratio is the reason this section exists. Search any of these and the results are videos telling you to raise a slider, and for nearly half of them there is no slider involved at all.