[Question] ImplicitActuatorCfg effort_limit and effor_limit_sim, differences and values due to motor parameters

[AndreaRossetto]

Question

I have this motor parameters:

(Nominal Voltage)	                       V	       48
(Voltage Range)	                       V	       24~56
(Power)                                           W            336
(Nominal Torque)	                        N.M	       7.5
(Stall Torque)	                                N.M	        22
(Nominal Speed after Reduce)	RPM	110
(Max Speed after Reduce)	        RPM	320
(Nominal Current)	                        A	        7
(Stall Current)	                                A	        22
(Phase Resistance)	                        Ω	        0.67
(Phase Inductance)	                        mH	        0.45
(Speed Constant)	                        rpm/v	6.67
(Torque Constant)                           N.M/A      1
(Number of Pole Pairs)	                Pairs	21
(Gear Rate)	                                 /	        8:1

I'm using the ImplicitActuatorCfg, is the correct one to simulate a gear reduced brushless motor?
What is the difference between effort_limit_sim and effort_limit?
The value for these two parameters is the Nominal Torque, the Stall Torque or something else?

[Bardreamaster]

For ImplicitActuator, effort_limit_sim and effort_limit are the same to perform a PD control law to clip effort in model like this, neithor Nominal Torque or Stall Torque. And the effort_limit will be removed in the future for ImplicitActuator. effort_limit_sim is for setting the effort limit implicitly computed by simulator. For more information, try have a look on changelog and source code

As I know, most brushless motors are DC motor, you might need this. ImplicitActuator should not be used directly , which is informed in docs.

[Nate711]

@Bardreamaster The docs say not to use ImplicitActuator but the explicit actuator is so unstable with medium to large damping values that even the example models use implicit actuators:

IsaacLab/source/isaaclab_assets/isaaclab_assets/robots/unitree.py

Line 217 in fe6ee74

"legs": ImplicitActuatorCfg(

. Given that it's impractical to use the explicit actuators because of the instability, what's your recommendation?

[Bardreamaster]

@Nate711 I'm not sure why the ImplicitActuator is still in examples.

For the question:

I'm using the ImplicitActuatorCfg, is the correct one to simulate a gear reduced brushless motor?

I think it's a good choice to use DC Motor Actuator.

And if you want to simulate all features of the motor you are using in real, data-driven sim motor might be good cause you do not have to have the dynamic model of the motor in simulator. eg: lstm motor. But you need more data to fit all operating conditions of the motor.

It would be better if designers give more information about motors simulator.

[AndreaRossetto]

Finally, I'm able to use the dc motor but now I've a new question about this. How precise is the torque simulation of the DCMotor? Let me explain better, I've created the robot model and I put the real components weights (cad estimated weight related to the material type) and real motor parameters, so if I simulate the model with this values and the simulated motor can handle the weights and they works fine, can I assume that also the real one will work? (in the model simulation I also add rigid bodies mass increase)

        "legs": DCMotorCfg(
            joint_names_expr = [ "hip_abduction_left", "hip_rotation_left", "hip_flexion_left",
                                 "hip_abduction_right", "hip_rotation_right", "hip_flexion_right",
                               ],
            saturation_effort   = 20.0,     # N-m ( Peak motor effort limit ),
            effort_limit        = 7.5,      # N.m ( Nominal motor effort limit ),
            effort_limit_sim    = 7.5,      # N.m
            velocity_limit      = 33.0,     # rad/s
            velocity_limit_sim  = 35.0,     # rad/s
            stiffness = {
                "hip_abduction_left":       1000.0,
                "hip_rotation_left":        1000.0,
                "hip_flexion_left":         1000.0,

                "hip_abduction_right":      1000.0,
                "hip_rotation_right":       1000.0,
                "hip_flexion_right":        1000.0,
            },
            damping = {
                "hip_abduction_left":       0.050,
                "hip_rotation_left":        0.050,
                "hip_flexion_left":         0.050,

                "hip_abduction_right":      0.050,
                "hip_rotation_right":       0.050,
                "hip_flexion_right":        0.050,
            },
            friction = 0.02,
            armature = 5.668e-4,
        ),

[RandomOakForest]

Thank you for following up. I'll move this to our Discussions section for further follow-ups.