Motor runs 25% slower woth Odrive S1 than with a average ESC controller

Support ODrive S1 Support
1

I use the Odrive S1 to test custom motors i make, i’m currently working on a 4KV motor, quite big and powerfull.
The issue is that with the same power supply set to 24 V and 10A the motor runs at a maximum of 1.2 rps when controlled by a ESC (sensorless and with no way to configure anything) and when controlled by the Odrive i can reach at maximum .9 rps.

Is that expected behavior or might i be configuring it wrong? I had to calculate the specs of the motor myself so i tried playing around with the KV when configuring the motor, but to no efect, the maximum is always 0.9rps

Also its the first time i have to configure a motor as gimbal! The phase resistance is 1.5ohm, i dont know if that might be related

2

Hi! Yes, the S1 has a 78% modulation depth limit, this is effectively the maximum utilizable percentage of DC bus voltage. This limit will be lifted in a future firmware version.

However, the gimbal mode may be also exacerbating this. You can try running in the HIGH_CURRENT/PMSM_CURRENT_CONTROL mode, or you can enable the wL_FF_enable feedforward term to help compensate for motor inductance. The bEMF_FF_enable term may also work, but it relies on an accurate torque_constant value (which is 8.27/kv, so it relies on an accurate KV value). Happy to give instructions for characterizing the kv experimentally!

1 Like
3

Thanks a lot for all the information, i’ll test all of it!
I’d really appreciate those instructions if you wouldnt mind!

4

For now i’m calculating kv like this: i have a comercial motor conected to an odrive spining at a set speed. The comercial motor is attached to the rotor of the tested motor. And with a multimeter i’m reading the voltage generated across two of the phases of the tested motor

1000095050
10000950501331×1080 317 KB

5

The “proper” way here is with an oscilloscope, Oskar showed the methodology here: Project HoverArm - #2 by madcowswe

6

Very cool post, thanks! I’ll look into getting an osciloscope to test it. Would you expect the results from both methods to stray a lot from each other?

7

Typically scoping it is the most accurate option, but zooming in I see that’s a true RMS multimeter, so what you had before might work fine!

Looks like (zooming in on the GUI screen) you’re spinning at about 4.7 rev/s?

68.2VAC (from the DMM) == 96.44 Vpk. So KV here would be (4.7*60)/96.44 = 2.92 RPM/V.

This seems pretty low, all things considered.

One other way you can do this:

  1. Setup/configure/calibrate the motor from the ODrive, you could even use the one mounted already and just use the existing magnet mounting, but run wires to your custom motor phase.
  2. Put the motor in velocity control, tune the gains to be stable.
  3. Spin the motor at a few different velocity setpoints ranging from about 25-75% of motor’s max free speed.
  4. At each velocity setpoint, measure axis0.vel_estimate and axis0.motor.foc.v_current_control_integral_q.
  5. KV will be vel_estimate * 60 / (sqrt(3) * v_current_control_integral_q)

This will be slightly inaccurate due to load on the motor and other losses, but it’ll get you a general ballpark. In a future firmware update, we’ll have the ability to automatically measure/characterize the motor KT :slight_smile: