Forums

Pages: 1 2

NaN Wrote:Cool. I'll have a look.

After endless tries i succeded in obtaining a somewhat working clutch.
This is the im using:

Code:
public float GetDragImpulse(float engine_speed, float drive_speed, float engineInertia, float drivetrainInertia, float ratio, float engineTorquedt){

    float drivetrainInertiaR=drivetrainInertia/ratio;

    impulseLimit=clutch_position*torqueCapacity*Time.deltaTime;

    float lambda=(engineInertia*drivetrainInertiaR*(drive_speed - engine_speed) - drivetrainInertiaR*engineTorquedt)/(engineInertia + drivetrainInertiaR);

    lambda = Mathf.Clamp(lambda, -impulseLimit, impulseLimit);

    return lambda;

}

engineTorquedt = engineTorque*Time.deltaTime;
drivetrainInertia=1;
engineInertia=0.5f;

drivetrainInertia is summed to each wheel inertia (1.8f).
This works with 50Hz and 1 single iteration.

Im still experiencing that infamous clutch lag, so I had to use a patch in order to decrease ratio value in drivetrainInertia/ratio by a value inverse proportional to the same ratio.

Let me know what you think.

No idea how to do it make it work. It is comparable to try running a rigid body simulation with a single iteration. Maybe there is a way to fake the clutch by scaling engine torque or something like that.

NaN Wrote:No idea how to do it make it work. It is comparable to try running a rigid body simulation with a single iteration. Maybe there is a way to fake the clutch by scaling engine torque or something like that.

Hi NaN eventually i make it work in a satisfying way by subtracting wheel impulse to clutch drag impulse calculation:

Code:
float lambda=(engineInertia*drivetrainInertiaR*(engine_speed - drive_speed) - x + drivetrainInertiaR*engineTorquedt)/(engineInertia + drivetrainInertiaR);

where x is wheelImpulse*engineInertia/ratio

in this way simulation runs satisfactory with 50Hz and 1 iteration (and its playable down to 20Hz). Soon i'll post a link for the demo.

Nice job. Sorry for the late reply, busy week.

Pages: 1 2

NewLife

NaN

NewLife

NaN