User guide
218
6000 Series Programmer's Guide
Master Position
Prediction
The master position prediction mode may be enabled or disabled with the FPPEN command,
but each state contributes a different error.
Disabled (FPPENØ): The slave position command is based on a master position that is 2
sample periods old. This means that master measurement error due to disabling the
position prediction mode will be (2 sample periods ∗ master speed).
Enabled (FPPEN1): If the position prediction mode is enabled (default setting), its accuracy is
also affected by position sampling accuracy, master speed, and master position
filtering. The error due to enabling the position prediction mode is about twice that
due to sampling accuracy (i.e., 40 to 200 microseconds ∗ master speed). As with the
error due to position sampling accuracy, the error due to the position prediction mode
being enabled is like a noise on the order of 200-600 Hz, which is not noticed by
large loads.
Master Velocity
Relative to
Master Position
Prediction &
Master Position
Filtering
Variation in Master Velocity: Although increasing master position filtering (increasing the
FFILT command value) eliminates the error due to sampling accuracy, it increases
the error due to variations in master speed when the master position prediction
mode is enabled (FPPEN1).
Most applications maintain a constant master speed, or change very slowly, so this
effect is minimal. But if the master is changing rapidly, there may be a significant
master speed measurement error. Because predicted master positions are in part
based on master speed measurement, the can result in an error in master position
prediction mode (FPPEN1). This effect will always be smaller than that due to the
master position prediction mode being disabled (FPPENØ).
Phase Tracking : The cost of using the low pass filter (increasing the FFILT command
value) is the increase in phase tracking error of the slave's position command. This
is an intrinsic characteristic of a low-pass filter. Increasing the filtering (lowering
the bandwidth) increases the phase tracking error, or phase lag.
If the master axis is moving at constant velocity, then the delay in the filtered
position in terms of time is
2
bw
sec , where “bw” is the bandwidth in radians/sec.
Therefore, at constant velocity, the tracking error of the filtered master position is
velocity
∗
2
bw
.
For example, suppose the master is moving at 4000 counts/sec and the master filter
bandwidth is 80 Hz, then the filtered master position delay =
2
80 ∗ 2π
= 3.98 msec,
and the slave position command tracking error = 4000 ∗ 0.00398 = 15.915 counts.
One important note is that the slave tracking error discussed here refers to the error
of the slave's position command calculated from the history of the master
position signal. The actual slave tracking error also depends on the tuning and the
dynamics of the system. For example, one way to eliminate the tracking error
introduced by the low pass filter is to use the velocity feedforward gain (SGVF) in
the servo loop.
Tuning
(Servos Only)
A servo system's tuning has a direct impact on how well the slave can track the master input.
Overshoot, lag, oscillation, etc., can be devastating to Following performance. The best tool
to use for tuning the 6000 series controller is Motion Architect's Servo Tuner Module. For
tuning instructions, refer to the Servo Tuner User Guide or to the Tuning
chapter/appendix your servo controller's installation guide.
Dynamic
Position
Maintenance
(Steppers Only)
Even when the slave is in motor step mode (ENCØ), there may be one slave step of error
inherent to the algorithm. When the slave is in encoder step mode, the user specified position
maintenance gain is used to correct position. This gain is expressed as motor steps per second
per encoder step error, and has a maximum value of 250 (limited internally). If a value lower
than this is used, the position error in encoder steps due to low gain is given by: error =
(250/gain) * (encoder resolution/motor resolution).