Gemini GV6K and Gemini GT6K Programmer's Guide

Master Position Prediction
Master Position Prediction is a technique used to compensate for the fact a follower's position
command cannot be calculated and implemented infinitely fast.
The master position prediction mode is enabled by default (FPPEN1) in the Following
algorithm, but can be turned off as desired with the FPPENØ command.
The Gem6K measures master position once per position sampling period (2 ms), and
calculates a corresponding follower position command. This calculation and achieving the
subsequent follower commanded position requires 2 sample periods (4 ms).
If master position prediction mode is disabled (FPPENØ), waiting 2 sample periods results in
a follower position lag. That is, by the time the follower reaches the position that corresponds
to the sampled master position, 2 sample periods have gone by, and the master may be at a
new position. Measured in time, the lag is 2 sample periods. Measured in position, the lag is
2 sample periods ∗ current follower velocity.
For example, suppose the follower is traveling at a speed of 25000 counts per second. If
master position prediction mode is disabled (FPPENØ), the follower will lag the master by
100 counts (25000 counts/sec ∗ 4 ms = 100 counts).
By measuring the change in master position over sequential sample periods, the master's
present velocity is calculated. The present master velocity and position are used to predict
future master position. If master position prediction mode is enabled (FPPEN1, the predicted
future master position is used to determine the follower's position command. In this case the
follower has no velocity-dependent phase delay. The follower's velocity for a given sample
will always be the velocity required to move from its current position to the next calculated
position command.
If the master motion is fairly smooth and velocity is not very slow, the measurement of its recent
velocity will be very accurate, and a good way of predicting future position. But the master
motion may be rough, or the measurements may be inaccurate if there is no filtering (see Master
Position Filtering below). In this case, the predicted master position and the corresponding
follower position command will have some error, which may vary in sign and magnitude from
one sample to the next. This random variation in follower position command error results in
rough motion. The problem is particularly pronounced if there is vibration on the master.
It may be desirable to disable the master position prediction mode (FPPENØ) when maximum
follower smoothness is important and minor phase delays can be accommodated.
If master filtering is enabled (FFILT_Ø), then the prediction algorithm would be used on the
filtered master position, resulting in a smoother follower position command. However, due to
the delay introduced by the filtering, the prediction algorithm would not compensate for the
total delay in the follower's tracking command. (See also Master Position Filtering below.)
Following Status (TFSF, TFS, and FS) bit 17 indicates the status of where or not the master
position prediction mode is enabled.
Master Position Filtering
The follower axis' position command is calculated at each position sample period (2 ms). This
calculation is a function of the master position and the master velocity estimated from the
change in master position over 2 position sample periods.
The Master Position Filter feature allows you to apply a low-pass filter to the measurement of
master position. Master position filtering is used in these situations:
• Measurement of master position is contaminated by either electrical noise (when analog
input is the master) or mechanical vibration.
• Measurement noise is minimal, but the motion that occurs on the master input is
oscillatory. In this case, using the filter can prevent the oscillatory signal from
propagating into the follower axis (i.e., ensuring smoother motion on the follower axis).
Chapter 7. Multi-Tasking 187