Follow the packet from the motion controller to each encoder and back: the master sends the cyclic request, EtherCAT slave hardware processes the frame, and each encoder returns actual position and status. Commission the physical measurement point first, then the network, distributed clocks, position references, mixed-axis control, and finally the complete machine.
Where should each encoder measure the motion?
Mount the feedback device where the controlled movement actually occurs. A motor-mounted encoder measures motor rotation, but gears, couplings, belts, drums, and compliant transmission elements remain outside that feedback path. Backlash, torsional deflection, slip, and lost motion can therefore produce load-position error even when motor position is correct.
| Measurement point | Position represented | Recurring limitation |
|---|---|---|
| Servo motor | Motor-shaft position | Does not directly detect downstream backlash or elasticity |
| Gearbox output | Reduced-speed output position | Requires a mechanically sound encoder coupling |
| Drum or take-up mechanism | Driven mechanism position | Does not detect material slip unless the material itself is measured |
| Measuring wheel | Linear travel inferred from wheel rotation | Wheel slip and diameter error affect the result |
For synchronized axes, place a mechanical absolute multi-turn EtherCAT encoder at the low-speed, post-transmission point when load position is the controlled variable. This avoids reconstructing multi-turn position from an incremental pulse counter and preserves a readable absolute reference after power interruption. It does not make the network immune to noise; it removes electronic turn counting as the stored-position dependency.
Check: Move the mechanism through its working range and confirm that encoder direction, scale, and displayed travel match the actual load rather than only the motor shaft.
How should the physical EtherCAT path be connected?
Layer one first. Trace the path from the master Ethernet port through every slave and cable segment. EtherCAT uses a master-and-slave bus structure with full-duplex Ethernet communication and logical frame circulation through the slaves. A failed link, reversed port assignment, unpowered slave, or defective connector stops the useful data path before controller logic can correct anything.
| Path element | Commissioning setting or inspection | Proof |
|---|---|---|
| Master interface | Select the controller port assigned to EtherCAT | The master detects the first slave |
| Slave order | Match the configured order to the wired order | Every encoder appears in the expected position |
| Cable and connector | Inspect seating, routing, shielding, and mechanical strain | No link interruptions occur while the machine moves |
| Branch | Keep branch devices and expected slave sequence aligned with the project | The branch enters cyclic operation with all slaves present |
| Optional ring | Add the master-side Ethernet path required by the selected redundancy design | A controlled cable interruption switches paths as designed |
EtherCAT supports line, branch, and ring arrangements, hot-connected network sections, cable redundancy, and hot-standby master designs. These functions require matching master capability and project configuration; a second cable alone does not create a validated redundant system.
Check: Confirm stable link indications and an exact match between discovered and configured slave order before enabling distributed clocks or motion.
Which cyclic data and timing settings must be established?
The controller needs each axis actual position and status in the same cyclic context used to calculate its next command. Map the encoder position into the process image, verify its data type and scaling, and confirm that the control task reads the new sample once per intended bus cycle. A fast network cannot correct an application task that reads stale data or executes asynchronously.
| Item | Reference figure | Commissioning use |
|---|---|---|
| 1,000 I/O update | 30 microseconds | Reference EtherCAT performance figure, not an automatic project cycle time |
| 100 servo-axis communication | Every 100 microseconds | Reference figure covering command, control, actual-position, and status exchange |
| Distributed-clock synchronization deviation | Less than 1 microsecond | Reference synchronization target to compare with measured project diagnostics |
Distributed clocks compensate slave-clock offsets across the network and provide a common time base for coordinated action and timestamped acquisition. They do not remove mechanical delay, controller-task jitter, or actuator response differences. Read the configured bus cycle, control-task period, clock state, and synchronization error from the master diagnostics rather than assigning the reference figures as acceptance limits without measurement.
Check: Trend consecutive samples from every encoder and confirm that positions update at the configured cycle with no stale, repeated, or unexpectedly skipped values.
How are direction, scale, and absolute zero commissioned?
A mechanical absolute multi-turn encoder supplies a unique encoded position across its supported mechanical range, but the controller still needs an axis coordinate system. Establish direction, engineering-unit scale, and the machine reference offset independently for every axis.
- Place the mechanism at a known mechanical reference using the machine commissioning method.
- Record the raw absolute position before applying offsets or modulo handling.
- Jog in the defined positive direction and verify that the raw value changes with the expected sign.
- Apply the conversion from encoder units to machine travel using the actual transmission relationship and measurement-wheel geometry where applicable.
- Set the controller’s reference offset so the known mechanical point reports the required machine coordinate.
- Cycle power without moving the mechanism, then compare the returned absolute position and calculated machine coordinate with the pre-power values.
Do not mask a direction error with a negative synchronization correction or conceal a scale error by repeatedly changing the zero offset. A zero adjustment corrects origin only; it cannot correct travel-per-count error, backlash, or wheel slip.
Check: Test at the reference point and at a second measured position. Both must agree after a power cycle, proving zero retention and scale separately.
How should mixed actuator axes use the same position reference?
Servo motors, variable-frequency-drive motors, hydraulic cylinders, and hydraulic motors can participate in one synchronized system when the motion controller receives comparable load-position feedback from every axis. The EtherCAT encoder becomes the common measurement interface; the actuators do not need to share one motor or drive family.
| Actuator | Primary controlled output | Feedback pitfall |
|---|---|---|
| Servo axis | Fast motor command with load-position supervision or closure | Motor encoder can hide gearbox and transmission error |
| Variable-frequency-drive motor | Speed or torque command corrected from load position | Open-loop speed behavior can create following error under load changes |
| Hydraulic cylinder or motor | Valve command corrected from measured motion | Fluid compliance, deadband, and load-dependent response affect synchronization |
Use the synchronized load-position samples to calculate each axis error against the commanded trajectory or designated reference axis. Tune each actuator path for its own dynamics. Identical position feedback timing does not make a hydraulic axis respond like a servo axis, and aggressive correction can excite compliance or produce oscillation.
Check: Run each axis alone at reduced speed, verify stable direction and following-error behavior, then couple the axes only after every individual loop responds correctly.
How is end-to-end synchronization verified?
Prove the complete path under controlled conditions: master request, physical frame path, clock alignment, encoder sample, controller calculation, actuator command, and actual load motion.
- Start with motion inhibited and verify that all configured encoders enter cyclic operation without communication or clock diagnostics.
- Move one mechanism manually where permitted and confirm that only its expected position channel changes.
- Jog each powered axis independently at reduced speed and validate direction, scale, limits, and stable feedback.
- Command synchronized motion at reduced speed while trending commanded position, actual position, inter-axis error, slave state, and clock diagnostics.
- Repeat at representative speed and load, watching for error growth during acceleration, reversal, and load transitions.
- Stop and restore power without moving the mechanisms. Confirm that every absolute coordinate returns correctly without a homing search.
- If redundancy is configured, introduce the approved controlled link interruption and verify the documented switchover behavior before returning the machine to service.
Check: Complete a representative synchronized cycle, power restoration test, and configured fault-path test with all axes retaining correct coordinates and the master reporting stable cyclic communication.
FAQ
Why does a servo axis lose synchronization when its motor encoder looks correct?
The motor encoder cannot directly measure gearbox backlash, coupling compliance, drum movement, or downstream slip. Add load-side absolute feedback at the gearbox output, drum, measuring wheel, or other actual-motion point.
Why does an EtherCAT absolute encoder still need a zero offset?
The encoder reports an absolute mechanical code, while the controller needs a machine coordinate. Record the raw value at a known mechanical reference, apply the axis offset, and verify the result at a second position.
Why does distributed-clock synchronization not remove following error?
Distributed clocks align sampling and communication time; they do not remove actuator lag, hydraulic compliance, gearbox error, or control-task jitter. Trend commanded position, actual position, and inter-axis error to locate the remaining delay.
Why does the EtherCAT network fail before motion starts?
Trace the master port, cable path, slave power, connector seating, and configured slave order. Establish stable cyclic communication first; position mapping and motion logic cannot repair a broken physical path.
Why is one successful synchronized cycle not enough?
Repeat the cycle through acceleration, reversal, representative load, and power restoration. The final verification is a complete synchronized cycle followed by a power cycle in which every axis returns to the same absolute machine coordinate without homing.