SINAMICS S120 Print Mark Correction with SINA_POS FB284

SINAMICS S120 Print Mark Registration: Achieving 0.5 mm Repeatability with Probe Input and SINA_POS FB284

1. Problem Statement

A continuous label-film perforation station uses a SINAMICS S120 servo (CU320-2 DP) driving an index roller, with a mechanical perforation ram mounted on a separate mechanism. A registration sensor (contrast or color-mark detector, 24 V PNP) detects the printed mark on the film, and the operator wants the axis to stop with the perforation tip aligned to the mark edge. Target positional repeatability is 0.5 mm across cycles; reported field repeatability is 1-3 mm, which causes the perforation tip to drift off the printed image area.

Compensation is implemented on a S7-1500 CPU over PROFINET (IRT) using SINA_POS (FB284) and telegram 111. The PLC reads the mark input bit through the CU I/Os word at the end of the receive telegram and reads a position value over the same bus, then computes a positional correction for the next index cycle.

2. System Architecture and Latency Layers

Print-mark registration on a moving web is deterministic only when three steps are deterministic on every cycle: a trigger event (mark edge), a timestamp event (latch the encoder position at the trigger), and a compensation event (move to a fixed LU offset from the latched value). Each layer between the trigger and the axis contributes variance:

Layer	Step	Typical contribution at 1 m/s web speed
Mark sensor	Edge response time	20-500 µs → 20-500 µm
SINAMICS DI / probe latch	Current-controller cycle (125 µs)	~125 µm
SINAMICS position filter	p2534 smoothing time constant	1-8 ms → 1-8 mm
PROFINET IRT	Send-clock and phase jitter	0.3-1 ms → 0.3-1 mm
S7-1500 OB	OB1 / OB35 / OB61 cycle and FB runtime	1-2 ms → 1-2 mm
SINA_POS propagation	Control word → drive setpoint	0.5-1 ms → 0.5-1 mm

Stacked, expected swing is 3-12 mm at 1 m/s — close to the observed 1-3 mm. The fix is to perform timestamping at the drive firmware layer (current-controller cycle, ~125 µs) and reduce PROFINET propagation to a confirmation word, not a measurement word.

3. Root Cause: Smoothing Filter and Bus Jitter

The as-built configuration reads r2521 as the mark position reference. r2521 is the actual position value processed by the position controller after the actual-value smoothing time constant p2534. The closed loop deliberately smooths the input to reduce acoustic noise and following-error ripple on the analog setpoint; it is not a deterministic capture of where the encoder was when the mark fired.

When the PLC value is the smoothed controller input instead of the raw encoder position, two error sources combine:

1. Filter phase lag introduced by p2534; scales linearly with web speed
2. Time-stamp misalignment from PROFINET send-clock jitter and S7-1500 OB1 execution jitter (1-2 ms typical)

A 0.5 mm target at 1 m/s permits a 0.5 ms total error budget. There is no room for a 1-8 ms position filter plus a bus plus an OB inside that budget. The objective is to read the position at the source (drive firmware) and ship only the resulting offset to the PLC.

4. Capture Strategy 1: Read r2523, Not r2521

The first, lowest-effort change is to switch the read parameter from r2521 (smoothed actual) to r2523. r2523 is the actual value of the measuring-gear position with no closed-loop filtering — the direct encoder value scaled by the load-side measuring-gear factor (p2504, p2505).

Parameter	Meaning	Suitability for mark registration
r2521	Actual position value in closed-loop control (smoothed)	No — filter lag
r2522	Actual value at the smoothing filter output	No — setpoint-side delay
r2523	Actual value of measuring-gear position, no smoothing	Yes — direct reference for PLC compensation

4.1 Read Path on PROFINET

Telegram 111's receive position word is mapped to a fixed signal source in the drive firmware and cannot be re-mapped from telegram configuration alone. Extend the telegram with a supplementary data slot in TIA Portal: SINAMICS drive → Properties → Telegram configuration → Supplementary data → add one word with source = r2523. After compile and download the S7-1500 sees the unsmoothed value in the supplementary receive slot.

Verify the active SINAMICS firmware release against the current SINAMICS S120/S150 List Manual before commissioning: parameter numbers, scaling, and the receive position word's ownership in the loop may shift between firmware branches.

5. Capture Strategy 2: Drive-Level Probe Measurement (Recommended)

SINAMICS S120 supports two measurement probes per encoder that latch the encoder position on a configurable DI edge with current-controller time-base resolution. The probes are independent of position-controller smoothing and run at the same 125 µs time base as the current controller — roughly an order of magnitude faster than a 1 ms PROFINET IRT send clock.

5.1 Probe Subsystem Function

• Input selection: configure the DI on the Control Unit or on the Sensor Module that triggers the probe
• Edge selection: rising, falling, or both edges; single-shot or continuous
• Latch register: encoder value at the trigger edge, scaled to load-side linear units (LU)
• Status flags: probe-valid bits in the SINAMICS control/status word or in the extended telegram status

The probe is the preferred measurement source for any touch-trigger application: flying saw, registration mark, edge detector, and length measurement. Configurable parameters are grouped under the Measurement subtree in the List Manual; indexes depend on the active firmware. Confirm against the active List Manual entry (search measuring probe in Siemens Industry Online Support) before commissioning.

5.2 Reading Probe Values on PROFINET

The probe latched value appears as a parameter (commonly routed through the encoder-level probe subsystem, e.g. registers in the 47xx range depending on firmware). To bring the value into the S7-1500 cycle, add a supplementary slot in telegram 111 mapped to that parameter. The valid-measurement status bit can also be mapped into the receive status word so the PLC knows when a new latched value is available.

6. Capture Strategy 3: EPOS Passive Homing + MDI

SINAMICS S120 in basic-positioner (EPOS) mode supports passive homing triggered by a configurable event. With a probe wired as the homing switch, the drive resets its position reference to the latched encoder value the moment the mark fires. Compensation becomes a fixed LU offset that the drive itself applies through MDI (Manual Data Input) absolute positioning.

1. Configure EPOS active homing at start-up (one-time, encoder reference to a permanent home)
2. Configure passive homing on the probe input (per cycle, initiated by the mark sensor)
3. Set MDI mode with absolute setpoint = (computed offset to perforation target) every cycle
4. The drive moves from the freshly latched homing reference to the absolute target inside the same firmware cycle

This removes the PLC from the capture-to-actuate path. The S7-1500 sends the position setpoint and the stop command; the mark edge is consumed by the drive firmware. PROFINET jitter no longer enters the compensation error budget.

7. Wiring the Registration Sensor

Wire the sensor directly to the CU320-2 DI, not into a S7-1500 DI module. Three reasons:

• Drive-firmware latch has shorter, deterministic latency than PLC input-map copy
• Eliminates a single-point-of-failure mode during PROFINET phase-up and removal
• Allows the engineer to trace mark-edge vs encoder value inside Startdrive without involving the PLC

Wire	Terminal on CU320-2	Notes
+24 V supply	X122.7 / X132.7	Short-circuit protected; total sensor + encoder load ≤ 500 mA per group
Sensor output (PNP)	X122.1 (DI 0) or X132.x	Match DI configuration to PNP sourcing
Sensor 0 V	X122.8 / X132.8	Dedicated return, not chassis
Shield	Cabinet PE rail, one end only	Avoid earth-loop current through shield

Configure the DI for PNP sourcing input (default on CU320-2). If the sensor uses a push-pull output, set the corresponding input filter / sensor type to match — otherwise a static-high may latch a wrong probe state on every cycle.

8. Telegram 111 Layout and Supplementary Slots

Telegram 111 is the standard telegram for SINA_POS. Default receive direction: status word 1, status word 2, actual position, actual speed, actual torque, CU I/Os (last word). Default transmit direction: control word 1, control word 2, position setpoint, override, MDI additional setpoints.

Process data word	Direction	Standard content
1	RX (drive → PLC)	Status word 1 (ZSW1)
2	RX	Status word 2 (ZSW2)
3	RX	Actual position (loop value)
4	RX	Actual speed
5	RX	Actual torque
6	RX	CU I/Os (mark sensor bit)
1	TX (PLC → drive)	Control word 1 (STW1)
2	TX	Control word 2 (STW2)
3	TX	Position setpoint (LU)
4	TX	Velocity override
5	TX	MDI additional setpoints
6	TX	MDI mode / control bits

Place supplementary slots to add:

• Unsmoothed position reference (r2523)
• Probe latched actual value
• Probe valid bit, if not already in ZSW2

The SINA_POS basic-positioner documentation covers the modes carried by telegram 111. The exact receive-word mapping is firmware-dependent — verify against the active firmware's List Manual entry before commissioning.

9. SINA_POS (FB284) Implementation in TIA Portal

SINA_POS (FB284) is the standard Siemens function block for S120 basic-positioner operation from a S7-1500. The block encapsulates the handshake required to drive telegram 111 in modes 1 to 6 (jogging, MDI absolute, MDI relative, traversing blocks, homing, setup). The block documentation is bundled with the SINAMICS Blocks library inside the DriveLib package — open the FB Help inside TIA Portal for full pin assignment and control-word sequencing.

9.1 Pattern for Print-Mark Compensation

1. Issue SINA_POS with ModePos configured for the active compensation mode on cycle start
2. Read the probe latched position from the supplementary slot when its valid bit transitions true
3. Compute compensated setpoint = latched value + target offset (LU)
4. Re-issue SINA_POS with the MDI-absolute mode carrying the new setpoint
5. Trigger the stop on the perforation synchronism output before the setpoint is reached

9.2 Worked Example (LU Math)

Assume the load-side measuring gear is 1 LU = 1 µm. At each mark edge the drive latches probe_value = 124000 (124.000 mm). The wanted perforation tip is 142.000 mm after the mark. The compensated setpoint for the next MDI absolute move is:

Position_next = probe_value + offset_LU = 124000 + 18000 = 142000 (= 142.000 mm)

If the next mark latched at 124150 (50 µm higher), the next setpoint becomes 142150 — the perforation tip is shifted by 50 µm upstream to compensate exactly for the same drift the next mark sensed. This is the closed loop.

10. Commissioning Trace Procedure

1. Open Startdrive (preferred for TIA Portal engineering of CU320-2 with newer GSD) or STARTER for legacy GSD-based commissioning
2. Configure a long-term trace: encoder 1 actual position, position controller input, position controller output speed, probe latched value, DI state on the mark input
3. Trigger on the DI rising (or falling) edge matching the configured probe edge
4. Run 100 production cycles and observe the spread of the probe latched value
5. Cross-check the value the PLC records two OB1 cycles later against the trace to measure PROFINET propagation lag
6. Compute the histogram and 6 σ; compare against the 0.5 mm target

Trace metric	Target	Action if not met
Probe latched repeatability (6 σ over 100 cycles)	≤ 50 µm in LU	Mechanically isolate sensor mount; verify web tension; increase sensor debounce
PLC read lag vs latched	≤ 1 IRT send clock (1 ms)	Run SINA_POS in OB61 (fast OB); re-check IRT topology in TIA Portal
Position controller overshoot at mark	< 0.3 mm	Reduce Kp of position loop; pre-shape velocity profile with EPOS acceleration
Settling time at mark	< 2 cycles	Reduce approach velocity; verify friction-compensation model
Following error in steady-state	< 100 µm LU	Verify Kv factor; check load inertia vs motor inertia ratio

If the probe latched value still drifts more than 50 µm across cycles after the drive trace is clean, the cause is mechanical: encoder coupling slip, belt stretch, or sensor-mount deflection. Verify against a hardware reference (hand-index one revolution) before further controller tuning — premature tuning of the position loop masks mechanical faults.

11. Acceptance Test for 0.5 mm Repeatability

1. Run 100 production cycles at full web speed
2. Log latched position at probe edge and final perforation position measured by a non-contact laser micrometer or vision system over the perforation tip
3. Compute system repeatability as 6 σ and as maximum deviation across the 100 cycles
4. Verify both metrics are within 0.5 mm
5. Repeat at minimum web speed, at maximum web speed, and after a 30-minute warm-state test to confirm thermal stability
6. Repeat after a PROFINET phase-down / phase-up cycle to confirm robustness against bus re-initialization

12. Troubleshooting Matrix

Symptom	Likely cause	First check
Probe never fires	DI not wired into probe input; wrong sensor type configured	Run Startdrive trace on the DI; verify input filter matches sensor output (PNP/NPN)
Probe fires intermittently	Sensor bouncing or web flutter	Add debounce (100-500 µs typical); shroud sensor from reflected light; verify web tension stable
Deterministic 1 mm offset per cycle	Wrong LU scaling (p2502, p2504) or wrong measuring-gear ratio	Rotate the load exactly one revolution; compare encoder revolutions in LU vs mechanical revolution
Swing grows with web speed	Position-loop bandwidth too low	Re-tune Kp and pre-control; reduce p2534 actual-value smoothing only if no other dependency on it
Compensation overshoots every cycle	MDI setpoint origin set every cycle from the wrong reference	Switch MDI setpoint origin to the probe latched reference for the cycle, not the absolute home
Drift after warm restart	PROFINET phase-up shifted the IRT phase	Recheck PROFINET topology; force a fresh startup with explicit IRT phase assignment in TIA Portal
Slow drift over hours	Encoder coupling slip / wear	Field inspection; replace coupling; re-verify mechanical repeatability
Repeatable oscillation on every cycle	Mechanical resonance with controller gain	Reduce Kp; add a low-pass on speed setpoint; check filter cut-off vs structural mode
Rapid deceleration overshoots perforation target	Position controller overshoot at endpoint	Lower deceleration in MDI block; activate pre-control / feed-forward
Perforation stops working in burst mode	Homing reference lost between cycles	Verify SINA_POS control word sequencing; capture homing-state bit in trace

13. Frequently Asked Questions