What Communication Type is Used for Servo Drive System?

It depends on the system design. Sometimes one has to (still) use a dedicated net just for servos but that is old. Different fieldbuses and/or servo drive interfaces may support a partial frame; for example, one may only be able to set a control word which might be position or torque command while some may support a full parameter frame or command set which exposes a plethora of servo parameters and variations between the two extremes. Speed is also a consideration i.e. may only be fast enough to do motion control or fast enough to close a servo torque loops and/or fast enough to incorporate feedback devices e.g. force or range sensors (without wiring directly to the servo drive) - again, how fast? There's obviously a difference between a lumbering 100 horse motor and a low inertia voice coil motor. There's also the issue of how many servos (along with other I/O) can be handled at one time and whether or not safety I/O is (ideally) incorporated into the servo net.

There's also a dichotomy in basic control ideas i.e. embed servo loops entirely within the drive, within a motion controller or within the PLC or PC and, of course, somewhere in-between. Obviously, closing the loop at higher levels demands faster, more robust and more deterministic communication performance; it also demands faster and deterministic scan times in the controller: e.g., for precision work, closing torque loops at 30-40 kHz. Control networks that provide real time clock distribution and time-stamped I/O are helpful (in maximizing determinism) as well as control schemes that support interpolation. An interesting variation is distributed I/O slices that perform motion control and may even support smaller motors without a drive or with only an external power driver module.

The etherCAT which is a general purpose field bus which has been demonstrated to close the loop on a large number of servos concurrently which would seem to be going in right direction.