Position setpoint with CST

Close the position and velocity loops in Motorcortex and let the drive follow a target torque.

  3 minute read  

Motorcortex closes the position and velocity loops and sends the drive a target torque every cycle. Use it when the application must control torque itself, for example for impedance control, which only works in a torque mode.

Motorcortex Drive Position setpoint Position loop Velocity loop Torque loop Motor Target torque · 6071 Position actual 6064 · Velocity actual 606C

Paths are under root/AxesControl/actuatorControlLoops/actuatorControlLoopNN/, where NN is the axis number.

Set the drive mode

In Desk, write 10 10 10 10 10 10 to root/DriveLogic/driveMode, one value per axis, and save with Save parameter tree to server. See Drive logic for details.

The standard configuration has these already. Otherwise, map the entries in the drive (CiA402 drives) and link them with channel 0 (EtherCAT configuration). Paths starting with …/ are under the axis’s control loop:

Motorcortex parameter Drive entry …/motorTorqueTarget 6071:00rxPDO Target torque to the drive …/motorPositionActual 6064:00txPDO Position actual value from the drive …/motorVelocityActual 606C:00txPDO Velocity actual value from the drive
See the links in the ECAT tool

Each entry shows its production link to the control loop and, below it, its simulation link to the Simulator. The blue marks the simulation link.

ECAT tool: the target torque entry of axis 1 with its production link to the control loop and its simulation link to the Simulator
ECAT tool: the position actual value entry of axis 1 with its production link to the control loop and its simulation link to the Simulator
ECAT tool: the velocity actual value entry of axis 1 with its production link to the control loop and its simulation link to the Simulator

Set the transformations

The transformations turn drive units into rad, rad/s and Nm at the load. This is what happens to a position reading:

Reading one motor revolution: 20-bit encoder, 100:1 gearbox Drive 1310720 ticks 1048576 ticks Motor 6.283 rad Load 0.0628 rad = application 1 − offset 2 ÷ gain 3 ÷ 100 1 offset In drive units: 262144 ticks here. Homing sets it, so the axis reads its reference position there. 2 gain ticksPerRevolution × gainNum / gainDen = 1048576 × 1 / 2π = 166886 ticks per rad 3 gearbox gearboxLoadSide / gearboxMotorSide = 100 / 1 motor turns per load turn Targets go the other way: drive value = value × gain + offset. For an encoder that restarts at 0 every turn, enableSingleTurnCounter first adds the whole turns it counts, times ticksPerRevolution.
Set Parameter Example
Gearbox gearboxLoadSide, gearboxMotorSide 100, 1 for 100:1
Encoder positionTransformation/transducer/ticksPerRevolution 1048576 for 20 bits
Zero position positionTransformation/transducer/offset, in ticks set by homing
Velocity units velocityTransformation/transducer/ticksPerRevolution 1 if the drive reports rev/s
Torque units torqueTransformation/transducer/gainNum, gainDen 500, 1 for 2 Nm rated, drive in ‰

Then move the axis a known amount and compare with actuatorPositionActual. If it moves the wrong way, make gainNum negative.

Tune the controllers

  1. Velocity controller first (velocityController/…): set controlOutputMin and controlOutputMax to the motor’s torque range, set parameter/enable to 1, and raise proportionalGain until actuatorVelocityError stays small.
  2. Position controller (positionController/…): set parameter/enable to 1 and raise proportionalGain until actuatorPositionError stays small without overshoot.
  3. If a steady error remains, widen controlIntegrationMin and controlIntegrationMax, then add integratorGain. Save the parameter tree.

Both are PID controllers. Double-click an error in the Desk Parameters panel to plot it while you tune.

Check the result

Plot actuatorPositionTarget and actuatorPositionActual in Desk during a move: they should overlap, and the error should settle to zero when the axis stops.

Next: Step 3: configure the axes.