Application Note: Custom Delta Robot Motor Tuning
Application Note: Custom Delta Robot Motor Tuning
Application Note: Custom Delta Robot Motor Tuning
Application Overview
This application note provides an overview of steps and an example to tune the three arm
motors for a custom delta robot. Before using this application note, it is assumed that the
user has properly configured the delta robot in Hardware Configuration, homed the motors,
and is able to move the robot in Machine Coordinate System (MCS) properly. The main
goal of tuning the motors should be to minimize vibration (hardware vibrations, motor torque
ripples) since it is typically the limiting factor in getting higher throughput.
The main strategy has the following steps to tune the three arm axes:
Use SigmaWin+ to find the inertia ratio
The theta axis is much easier, and can typically be tuned like a regular
motor.
This is meant to be a general guide of possible steps, and the user may need to slightly
deviate from it depending on the needs of their mechanism and application.
Products Used
Delta Robot Any simple delta robot with kinematics supported by Hardware Configuration
Implementation
CAUTION
When using SigmaWin+ for inertia finding or auto tuning, the motor controlled by
SigmaWin+ will move as specified by the inputs to the SigmaWin+ routines. Ensure that
the mechanism will have enough space to not crash.
When doing manual tuning using a defined path, ensure that the application in
MotionWorksIEC commands a path that the mechanism can follow without crashing.
This tuning procedure assumes homing of each axis and is already done and delta
mechanism is configured properly in Hardware Configuration in MotionWorksIEC.
Yaskawa is not responsible for damange to mechanism or damage caused by mechanism
during this procedure.
Before starting, make sure the robot motors are in a good position to avoid obstacles or
collisions. The inertia ratio finding and auto tuning routine typically require less than two
motor revolutions of movement. Typically, motor positions of 0 degrees is a good starting
point. Open SigmaWin+, and connect to S L or U motor
Pn109 Feed Forward Gain – Setting to 100 may help for, especially with
1310 disabled
Pn506 Brake reference servo OFF delay time – When servo off is
commanded, keep power on for some time until brake is engaged. Set to
150ms
Edit Pn520 to some large value. Default is 5242880 units, which should be
fine for a SigmaV motor. For a Sigma7 motor, a larger value like 15000000
units may be necessary. Select the Write Edited Parameters From the
SERVOPACK to save the new value.
Edit the Condition Settings of the inertia ratio finding routine. Typically a
low Acceleration and Speed has a higher chance of completion. A higher
than default value (400) of Pn100 may also need to be used. Click Next to
go to Reference Transmission.
Autotuning
After finding an inertia ratio, the autotuning function in SigmaWin+ provides a good starting
point for motor gains and tuning parameters.
Autotuning may be aborted due to some possible causes, and there may be
an alarm on the drive. Deal with and clear the alarm. It may be necessary to
increase the value of Pn520 (Excessive Position Error Alarm Level), Pn522
(Position Completed Width), or Pn402/403 (Torque limits). In this example,
Pn520 = 15000000 and Pn522 = 1000 was used for a Sigma7 motor.
Pn520 = 5242880 and Pn522 = 70 be used for a Sigma5 motor. Other
values may work as well.
Restart the autotuning routine with the same settings as before. Allow the
tuning routine to complete. Click Finish to exit. The tuning parameters are
now saved.
Copy the parameters of this arm motor into the other 2 arm motors. Can do
it though SigmaWin+ or MotionWorksIEC.
Manual Tuning
After finishing autotuning, some manual tuning may be required to get rid of excess
vibration or torque ripples.
Open SigmaWin+ and connect to an arm motor. Go into the Trace menu.
The plot may look something like this with some torque ripple.
FIGURE 19 FEEDBACK SPEED, TORQUE REFERENCE, POSITION ERROR WITH AUTOTUNE PARAMETERS
Manually tune parameters in the Edit Parameters menu until satisfied with
amount of torque ripple and position error. Typically, editing Pn100, Pn101,
and Pn102 will be good enough. Keep in mind that the first priority is
minimizing vibration, which may shows up as torque ripples. High speed
video of the TCP would also be useful to ensure the payload is not vibrating
during moves. For Sigma5 servopacks, controller parameter 1310 may
need to be disabled. After tuning, the plots should show a smoother torque
profile and TCP vibration should be minimized.
FIGURE 20 FEEDBACK SPEED, TORQUE REFERENCE, POSITION ERROR AFTER MANUAL TUNING
Stop the move routine, save the parameters, and copy the parameters over
to the other two arm axes.
Run sample path again and check SigmaWin+ on all arm axes to make sure
that the tuning is still satisfactory.
Sample plot of an arm axis when running the sample path at high speed.
7000mm/sec velocity, 120000mm/sec^2 acceleration after manual tuning. In
this data set, the max position error after converting to arm angle is about
0.45 degrees. (around 800000 units max position error, 24bit encoder for
Sigma7 motor, 38.5 gear ratio)
It is calculated as:
Run a sample path at a slow speed. The same path from manual tuning was
used.
Run sample path at high speed, and plot Torque FeedForward, Torque Reference, and
Position Error, then compare to the previous plot before Torque Passthrough was
enabled. From the following plot, it shows that when Torque Passthrough was enabled,
Position Error has dropped significantly, from a maximum of 800,000 (after manual tuning)
units to 500,000 units.