What Effect Does Gain Have on Performance?
The higher the gain, the less error (E) required to break friction or maintain velocity. The error required to
break friction will affect position accuracy at the end of a move, which makes it a major factor in
achieving repeatability. The error to break static friction can be measured with the loop closed by slowly
changing the command (C) by its least increment while observing the buildup of the error (E). As noted
earlier, a velocity loop will have a major impact on the error required to break friction. This test should be
done at several points along the travel since mechanical variations will cause the breakaway friction to
change.

Another common problem is null hunt, a phenomenon in which an axis moves back and forth with a
square waveform at a low frequency. This is usually caused by the breakaway or static friction being
significantly higher than the running friction. Essentially, the error builds up to break friction, but once
motion starts the error is more than necessary to maintain the desired velocity so it overshoots the desired
position. This continues to repeat in both directions. It can be prevented by lowering the gain, however
lowering the gain will also affect accuracy. Lowering the ratio of static to running friction can be
achieved with roller bearings or, as is more common now, through the use of a special coating material as
one of the bearing surfaces. A static to running ratio of 1.01 or less is achievable in this manner.

Accuracy during motion is a concern in many applications. Cutting metal, routing wood, etching glass,
and grinding silicon wafer edges are examples where extreme accuracy during motion is required. A
servo with a gain of 1 IPM/MIL will have 0.001" of error when traveling at 1 IPM, 0.01" at 10 IPM and
0.1" at 100 IPM. It follows that the best accuracy can be achieved by keeping velocities low and gain
high. This is a good generality, but not always that simple to achieve.