S7-300 Encoder Pulse Timing: Measure Period with OB40 Counters

1. Problem Definition: Reject Position Drift on a 36000 BPH PET Line

A PET bottling line running at 36000 bottles/hour produces one bottle every 100 ms (3,600,000 ms per hour divided by 36,000 = 100 ms per bottle). The label inspection and reject system must:

• Track each bottle's position via an encoder (10 pulses/revolution)
• Build a bit-shift register in OB1 that follows the bottle from the inspection point to the reject point
• Actuate a reject cylinder exactly when the bottle is in front of the rejecter

Reported symptom: the rejecter sometimes pushes a bottle that is "a little behind" or "a little late" of the expected position. The pulse count between bottles is not constant because the gap between two encoder pulses is "bigger than required" and the "conveyor speed also varies."

Fundamental timing constraint at 36000 BPH:

• 1 bottle = 100 ms window
• CPU 317-2 PN/DP (6ES7317-2EK14-0AB0) typical OB1 scan = 3 ms (as reported)
• Encoder disc with N pulses per bottle yields pulse period = 100 ms / N

If a pulse high time drops below the OB1 cycle plus the SM 321 input filter (~3 ms default), the discrete input is read only on alternate OB1 passes. A 6.25 ms pulse period sampled at 3 ms means roughly half of the OB1 cycles will see the pulse high and half will not, depending on phase relationship. This is the root cause of the rejection drift.

2. System Architecture and Hardware Part Numbers

Reference: SIMATIC S7-300 CPU 31xC and CPU 31x Operating Instructions (Siemens support entry ID 12996906), and the S7-300 Module Data Reference Manual (8859629).

3. Timing Analysis: Why Pulses Are Lost

Apply sampling theory to discrete inputs in OB1:

• Sampling rate: f_sample = 1 / T_OB1
• Signal rate: f_signal = 1 / T_pulse
• A pulse is detected when its high time T_high ≥ T_OB1 + T_{input_filter}

For the user's application:

• T_OB1 ≈ 3 ms (measured)
• T_high ≈ T_low = T_pulse / 2 (50% duty cycle, square wave)
• For N = 16 teeth: T_high = 3.125 ms, T_low = 3.125 ms
• SM 321 input filter default 3 ms: T_high barely exceeds the filter; jitter at the input may suppress the pulse entirely on some scans.

4. Solution A: OB40 Hardware Interrupt with Timestamp (Software-Only on Interrupt-Capable DI)

SIMATIC S7-300 supports hardware interrupts on digital inputs when the input is configured to OB40. Each event records:

• OB40_POINT_ADDR: byte/bit address of the triggering input
• OB40_TIMESTAMP: 100 ns units timestamp of the interrupt (CPU 31x-2 PN/DP V3.3+)

OB40 runs in microseconds, decoupled from OB1, capturing the pulse edge deterministically. The bit-shift register pointer is incremented inside OB40, eliminating OB1 jitter.

4.1 Configuration in STEP 7 / TIA Portal

1. Open the SM 321 DI properties in HW Config (TIA: Device view > DI module > Properties)
2. Enable "Hardware Interrupt" on the encoder input channel
3. Assign OB40 to the rising-edge trigger (select "Rising edge" to avoid double-counting on falling edge)
4. Set input filter to 0.1 ms (or smallest supported)
5. Download hardware configuration to the CPU

4.2 Sample OB40 in STL

OB40 // Hardware interrupt OB - runs on encoder rising edge
      L     #OB40_TIMESTAMP        // 100 ns ticks, DWORD
      T     "db_edge_time".time    // store as last edge time
      L     #OB40_TIMESTAMP
      L     "db_edge_time".prev
      -D
      T     "db_edge_period"        // period in 100 ns units
      L     "db_edge_time".time
      T     "db_edge_time".prev     // shift for next call

// Increment bit-shift register write pointer (circular)
      L     "bsr_write_ptr"
      L     1
      +I
      T     "bsr_write_ptr"
      L     "bsr_max_len"
      MOD
      T     "bsr_write_ptr"
      SET
      =     "bsr_data".bit["bsr_write_ptr"]   // mark bottle present at write position
      CLR
      =     "bsr_data".bit["bsr_write_ptr"]   // (illustration only - DB bool array access syntax varies by tool)

OB40 timestamp resolution: 100 ns on CPU 317-2 PN/DP (firmware V3.3+). Use SFC64 "TIME_TCK" to correlate OB40 timestamps to wall clock and to read system time for logging.

Reference: SIMATIC S7-300/400 System Software, System and Standard Functions Reference Manual, OB40 and SFC64 chapters (44240604).

5. Solution B: FM 350-1 Counter Module with SFB47 "COUNT"

The FM 350-1 (6ES7350-1AH03-0AB0) provides a dedicated counter independent of OB1:

5.1 Configuration in HW Config

1. Insert FM 350-1 in slot 4 (right of CPU, before signal modules)
2. Set input type to 24 V single-ended (for typical PNP encoder) or 5 V differential (for RS422)
3. Select "Count continuously" mode
4. Enable SW gate (CPU controls counting start/stop)
5. Enable HW interrupt on comparator match and on zero/overflow
6. Assign OB40 to comparator interrupt
7. Download configuration

5.2 Wiring

• Encoder A → FM 350-1 pin 1 (signal A)
• Encoder 24 V → FM 350-1 pin 3 (encoder supply)
• Encoder GND → FM 350-1 pin 4 (M)
• Encoder shield → FM 350-1 pin 5 (functional earth)

5.3 SFB47 "COUNT" Parameters

5.4 Sample STL: Start the Counter and Read the Value

// First scan: configure FM 350-1, leave gate closed
AN    "m_first_scan_done"
S     "m_count_run"
R     "m_first_scan"
CALL  "COUNT" , "DB_COUNT"
  LADDR   :=256          // FM 350-1 base address 100h
  CHANNEL :=0
  SW_GATE :="m_count_run"
  CTRL_DO :=
  SET_DO  :=
  JOB_REQ :=0
  JOB_ID  :=0
  JOB_VAL :=
  STS_GATE:=
  STS_DO  :=
  STS_C_DN:=
  COUNTVAL:="md_count_val"
S     "m_first_scan_done"

// Each scan: read current count
CALL  "COUNT" , "DB_COUNT"
  LADDR   :=256
  CHANNEL :=0
  SW_GATE :="m_count_run"
  COUNTVAL:="md_count_val"

Reference: SIMATIC FM 350-1 Counter Module Function Manual, Siemens support entry 1404537.

6. Solution C: Pure Software Pulse Period Measurement with SFC64 TIME_TCK

When hardware changes are not permitted, measure the time between two rising edges using SFC64 "TIME_TCK" in OB1. The CPU 317-2 PN/DP does not timestamp discrete inputs natively, so the technique reads the system clock each time a pulse-edge is observed in OB1 and subtracts the previous reading.

SFC64 "TIME_TCK" returns the CPU system time as a DWORD. When the operand passed is type DWORD, the resolution is 10 ms; when the operand is type INT, the resolution is 1 ms. Use the INT variant for finer resolution.

// SFC64 call to read system time in 1 ms units
CALL  "TIME_TCK"  // SFC64
  RET_VAL := "mw_clock_ms"   // WORD, 1 ms units, wraps at 32767 ms

// Edge detection for encoder input I 0.0 (example)
A     "i_encoder"
FP    "m_encoder_fp"   // one-shot on rising edge
JC    _meas

_meas: L     "mw_clock_ms"
      T     "mw_clock_ms_prev"  // save for next cycle
      L     "mw_clock_ms"
      L     "mw_clock_ms_prev"
      -I
      T     "mw_period_ms"      // period in ms

// Exponential moving average: avg = (avg * 7 + period) / 8
      L     "md_avg_period"
      L     7
      *D
      L     "mw_period_ms"
      +D
      L     8
      /D
      T     "md_avg_period"

Accuracy is bounded by the OB1 cycle: a pulse edge can be observed 0 to T_OB1 (3 ms) after it occurs. For pulses whose high time is ≥ 3 ms, this technique yields a usable average within ±1 pulse. Missed pulses appear as period jumps of 2× expected; a 50% guard band on the average handles this:

// Reject distance in pulses
      L     "md_reject_distance_mm"  // physical distance inspection to reject
      L     "md_avg_period"          // average ms per pulse
      *D
      L     "md_conveyor_speed_mmps" // mm/s conveyor speed
      /D
      T     "mw_reject_pulses"       // integer number of pulses

7. Bit-Shift Register with Adaptive Reject Distance

The original bit-shift register uses a fixed reject distance (constant index offset from write pointer to read pointer). When the conveyor speeds up or slows down, the pulses-per-bottle ratio changes and the fixed offset drifts relative to the actual bottle position.

7.1 Adaptive Algorithm

1. Maintain bsr_write_ptr updated by OB40 (Solution A) or by FP-edge in OB1 (Solution C).
2. Maintain bsr_read_ptr = bsr_write_ptr - reject_distance_pulses (mod bsr_max_len).
3. Recalculate reject_distance_pulses each scan from the running average period and the known mechanical distance between the inspection sensor and the rejecter.
4. Set reject output when bit at bsr_read_ptr = 1 (bottle is at the rejecter).
5. Clear the bit at bsr_write_ptr immediately after writing if a single-pulse window is needed.

7.2 Minimum Stroke Time

// Minimum stroke stretch
A     "bsr_data".bit["bsr_read_ptr"]
=     "m_reject_raw"

A     "m_reject_raw"
L     "t_stroke_time"   // 80 ms
SD    "t_stretch_reject"
A     "t_stretch_reject"
=     "q_reject_cylinder"

8. Solution D: Proximity Sensor on a 16-Tooth Sprocket (Hardware Augmentation)

The most cost-effective mechanical fix is to add a proximity sensor (or photoelectric sensor) detecting teeth on a sprocket coupled to the conveyor. With 16 teeth at 36000 BPH:

• Pulse period = 100 ms / 16 = 6.25 ms
• Pulse high time ≈ 3.1 ms (50% duty) — marginal with a 3 ms input filter but adequate with a 0.1 ms filter
• Position resolution = 22.5 mechanical degrees per pulse

This raises the pulse rate above the OB1 sampling threshold (after filter adjustment) and gives finer position resolution. Pair this with Solution A (OB40) for full immunity to OB1 jitter.

9. Comparison of All Solutions

Recommended sequence for this application:

1. Implement Solution A first (lowest cost, highest payoff) by replacing SM 321 with the interrupt-capable variant.
2. If further precision is needed, layer Solution D (16-tooth sprocket + sensor) to multiply position resolution by 1.6×.
3. If pulse rates exceed 1 kHz or the system must count multiple encoders, upgrade to Solution B (FM 350-1).

10. Verification and Commissioning Procedure

1. Disable the reject cylinder (manual mode). Set m_test_mode = 1 to inject bottle detection at a fixed rate (10 BPH) for initial verification.
2. Trace OB40_TIMESTAMP in a data block or in the CPU online watch table. Confirm monotonic increments and that no two consecutive timestamps differ by more than 2× the expected period.
3. Connect the encoder input to a 1 Hz test pulse (function generator or PLC-built test pulse). Confirm OB40 fires exactly once per second with timestamp delta of 1,000,000,000 (100 ns units = 1 s).
4. Reconnect to the live encoder. Oscilloscope the encoder output and capture the bit-shift register pointer for 10,000 pulses into a circular trace buffer; verify no skipped increments.
5. Run the line at 36000 BPH with the reject cylinder disabled. Place a fluorescent-marked bottle upstream. Verify via vision or position sensor that the bit at the rejecter index is set to 1 when the marked bottle is in front of the rejecter.
6. Re-enable the reject. Run for 8 hours. Count reject actuations and visually inspect the reject log for false rejects or missed rejects. Target: 0 missed rejects, < 0.1% false rejects.
7. Check the CPU diagnostic buffer in STEP 7 (PLC > Diagnostic Buffer) for OB40 start events: one entry per pulse, no error codes.

10.1 Diagnostic Buffer Code Reference

11. Spare-Part Selection for Upgrades

All part numbers are orderable through the Siemens Industry Mall and are supported by TIA Portal V15.1 and later. The S7-300 family is in phase-out but remains in active spare-parts supply per the SIMATIC S7-300 product phase-out notice (109751706).

12. Field-Proven Caveats

• Mechanical slip: The largest source of position drift not solvable in software. Verify that the encoder is spring-loaded and tracks the belt, not the conveyor pulley. A worn or loose belt will cause the encoder count to diverge from the bottle position over a full conveyor run.
• VFD-driven conveyor: Variable-frequency drives change the speed within ±2% during acceleration and deceleration. The bit-shift register must tolerate a ±2 pulse variation in bottle spacing during line startup. A startup delay (timer 3-5 s) before enabling the reject prevents spurious rejects during ramp-up.
• Bottle skip or jam: If a bottle lies on its side, the optical sensor may not detect it. Use a backup proximity switch on the conveyor frame (separately wired) to detect "conveyor moving but no bottle at inspection" conditions and flag a line alarm.
• OB40 priority: OB40 runs at priority class 16 by default. Set it higher than OB1 (priority class 1) to ensure deterministic pulse handling. OB40 priority is configurable in HW Config under CPU > Properties > Priority.
• OB40 nesting: If the pulse rate exceeds the OB40 service time, OB40 will queue events. If the encoder produces 2000 pulses/s and the OB40 logic takes 100 µs, OB40 will run 20% of CPU time. Check CPU utilization with STEP 7: PLC > Module Information > Scan cycle time. Target OB1 cycle < 50% of available time.
• Power supply headroom: The interrupt-capable SM 321 (6ES7321-7BH01-0AB0) draws 30 mA more from the 24 V backplane than the standard SM 321. Check the PS 305/PS 307 load calculation when upgrading.
• Stepper / VFD interference: Encoder signals running parallel to VFD output cables pick up common-mode noise. Use shielded twisted-pair cable, ground the shield at the FM 350-1 end only, and route the encoder cable at least 30 cm from VFD cables.

13. Alternative Platforms (S7-1200 / S7-1500 Migration Path)

If the S7-300 is approaching end-of-life in the user's plant, the modern Siemens equivalents are:

The S7-1500 platform resolves the OB1 jitter problem at the CPU level (deterministic cycle), so the software-only path (Solution C) is fully adequate on S7-1500 for pulse rates up to 5 kHz. Migration effort is 1-3 engineering days for a system of this size.

14. Spare Logic Block: Adaptive Period FB in SCL

The following SCL function block (FB) implements an adaptive period filter that can be called from OB1 once per scan. It returns the current average pulse period in milliseconds and a flag indicating data validity.

FUNCTION_BLOCK FB_PulsePeriod
VAR_INPUT
  i_encoder : BOOL;          // raw encoder input
END_VAR
VAR_OUTPUT
  q_period_ms : INT;         // current period in ms (0 if not yet valid)
  q_avg_period_ms : INT;     // exponentially-filtered period in ms
  q_valid : BOOL;            // TRUE after at least 2 edges observed
END_VAR
VAR
  s_last_edge : INT;         // SFC64 reading at last edge
  s_now : INT;               // SFC64 reading now
  s_period : INT;            // raw period
  s_avg : INT := 0;          // running average
  s_count : INT := 0;        // edge count
  s_fp : BOOL;               // edge memory bit
END_VAR
BEGIN
  // Read system time in 1 ms units
  s_now := TIME_TCK();  // conceptual; SFC64 invocation in SCL uses a wrapper or direct call

  // Edge detection
  IF i_encoder AND NOT s_fp THEN
    // Rising edge detected
    IF s_count > 0 THEN
      s_period := s_now - s_last_edge;
      IF s_period < 0 THEN
        s_period := s_period + 32767;  // handle wrap-around
      END_IF;
      q_period_ms := s_period;
      // Exponential moving average, alpha = 1/8
      s_avg := (s_avg * 7 + s_period) / 8;
      q_avg_period_ms := s_avg;
      q_valid := TRUE;
    END_IF;
    s_last_edge := s_now;
    s_count := s_count + 1;
  END_IF;
  s_fp := i_encoder;
END_FUNCTION_BLOCK

Notes on SCL usage: SFC64 "TIME_TCK" is invoked via the SCL wrapper or via inline STL. The wrap-around handling assumes the 1 ms counter wraps at 32767 ms; if higher resolution is needed, use the DWORD variant (10 ms resolution, 49-day wrap).

15. Summary and Field Decision Tree

1. Confirm the line speed in BPH and the pulses-per-bottle ratio. Calculate pulse period. If pulse period < 5 ms, OB1 alone will miss pulses; proceed to step 2.
2. If you can swap the SM 321 to the interrupt-capable variant 6ES7321-7BH01-0AB0, configure OB40 on rising edge and capture timestamps. This is the lowest-cost, highest-payoff fix.
3. If the encoder runs faster than 1 kHz or you need absolute position recovery after power-off, install an FM 350-1 counter module and use SFB47 to read the count.
4. If hardware changes are forbidden, apply Solution C (SFC64 in OB1) with an exponential moving average and accept the ±1 pulse accuracy. Do not expect this to be deterministic; it is a best-effort estimate.
5. For all solutions, stretch the reject output to a minimum of 80 ms to ensure full cylinder stroke, and add a 2-3 second startup delay to ignore the conveyor ramp-up.
6. Validate against a fluorescent-marked bottle; do not trust the encoder count alone.

FAQ