Motor velocity actively limited for unknown reason

[maxkl_e]

I have an iPower GM5208-12 motor connected to an ODrive S1, with an AS5048A encoder. There is a plastic disk connected to the motor that I want to spin at a constant velocity, so practically no load on the motor.

The problem I encounter is that the ODrive seems to actively limit the velocity for a reason I do not understand. I expect it to be able to spin at k_v * v_dc * 0.78. The motor is rated at 24 rpm/V, so with 24 V it should reach 450 rpm, and at 30 V up to 560 rpm. However, the velocity saturates at around 420 rpm at 25 V and doesn’t increase further with increasing supply voltage. Setting the target velocity to 15 rps and slowly increasing the voltage from 20 V to 30 V yields the following behavior:

On an oscilloscope connected between two motor phases at the ODrive terminals I can observe that at 25 V, the motor voltage does indeed reach the maximum 78 % duty cycle during the sine wave peaks. However, at 30 V, the duty cycle only goes up to around 66 %. The effective voltage is 25 V * 78 % = 19.5 V and 30 V * 66 % = 19.8 V, so this matches the observed behavior.

The ODrive is configured with the following values (I believe these are all the relevant parameters):

# Power supply
print('Configuring power supply')
odrv0.config.dc_bus_overvoltage_trip_level = 31
odrv0.config.dc_bus_undervoltage_trip_level = 10.5
odrv0.config.dc_max_positive_current = 2
odrv0.config.dc_max_negative_current = -0.01
odrv0.config.brake_resistor0.resistance = 2
odrv0.config.brake_resistor0.enable = True
odrv0.clear_errors()

# Motor parameters
print('Configuring motor parameters')
odrv0.axis0.config.motor.motor_type = MotorType.PMSM_VOLTAGE_CONTROL # For a gimbal motor with high phase resistance
odrv0.axis0.config.motor.pole_pairs = 7
odrv0.axis0.config.motor.phase_resistance = 7.6
odrv0.axis0.config.motor.phase_resistance_valid = True
motor_kv = 24 # rpm/V
odrv0.axis0.config.motor.torque_constant = 8.27 / motor_kv
odrv0.axis0.config.motor.current_soft_max = 1.5
odrv0.axis0.config.motor.current_hard_max = 3
odrv0.axis0.config.motor.calibration_current = 0.5
odrv0.axis0.config.motor.resistance_calib_max_voltage = 4
odrv0.axis0.config.calibration_lockin.current = 0.5
# run_state(odrv0.axis0, AxisState.MOTOR_CALIBRATION) # apparently not necessary for MotorType.PMSM_VOLTAGE_CONTROL
odrv0.axis0.motor.motor_thermistor.config.enabled = False
#odrv0.save_configuration()

# Control mode
print('Configuring control mode')
odrv0.axis0.controller.config.control_mode = ControlMode.VELOCITY_CONTROL
odrv0.axis0.controller.config.input_mode = InputMode.VEL_RAMP
odrv0.axis0.controller.config.vel_limit = 15
odrv0.axis0.controller.config.vel_limit_tolerance = 1.3333333333333333
odrv0.axis0.config.torque_soft_min = -math.inf
odrv0.axis0.config.torque_soft_max = math.inf
odrv0.axis0.trap_traj.config.accel_limit = 10
odrv0.axis0.controller.config.vel_ramp_rate = 10

Does anyone have an idea if there are further configuration parameters or hardware limitations that can limit motor velocity?

[maxkl_e]

Here is a scope capture of the phase-to-phase voltage at 30 V, where the duty cycle only reaches 66 %:

[solomondg]

Hi! If you’re running in PMSM_VOLTAGE_CONTROL, then this is normal – that’s only really applicable for low-speed control, it won’t compensate for things like motor inductance (which is what prevents the higher speed control in this case). I’d recommend running in PMSM_CURRENT_CONTROL. You may have to manually set the phase inductance, since it’s higher than the typical acceptable range for the S1; I measured a GM5208-12 on my LCR meter, and got a number of 5mH. Try setting:

odrv0.axis0.config.motor.motor_type = MotorType.PMSM_CURRENT_CONTROL 
odrv0.axis0.config.motor.phase_inductance = 5e-3
odrv0.axis0.config.motor.phase_inductance_valid = True
odrv0.axis0.config.motor.phase_resistance = 7.6
odrv0.axis0.config.motor.phase_resistance_valid = True
odrv0.axis0.config.motor.current_control_bandwidth = 50
odrv0.save_configuration()

If you need to specifically run in voltage control mode, you can set axis0.config.motor.bEMF_FF_enable to True; note that this will require an accurate torque_constant. Measuring on my desk, I’m seeing a KV of more along the lines of ~40 RPM/V (thus Kt of more like 0.206 Nm/A). You can tune this by putting the motor in torque mode with a zero torque setpoint, then moving it by hand – if you feel a “drag” then the torque constant is too low, if you feel the motor trying to accelerate as you move it, the torque constant is too high.

[maxkl_e]

Thank you very much, that did indeed increase the maximum speed to around 500 rpm at 30 V. In what instances would I need to run in voltage control mode? I couldn’t really notice a difference with bEMF_FF_enable set to True or False in my quick test.

One issue I have in this mode is that at zero/slow speed, the motor moves quite a bit back and forth in a seemingly random way. I assume this is is because of limited resolution of the current measurement, am I correct? Is there any way to mitigate that?

Re KV: I determined the value of 24 rpm/V myself by spinning the motor with an electric screwdriver and observing the open-circuit phase-to-phase voltage on an oscilloscope. I got an amplitude of 5.9 V at 138 rpm, so 23.4 rpm/V. This also matches the manufacturer spec for an unloaded motor of 456-504 rpm at 20 V. How did you determine the KV?