S7-300 PLC Setpoint Selection with Hysteresis for 6 Fan Outputs

Problem Definition: Dual-Source Setpoint Selection for Six Cooling Fans

A SIMATIC S7-300 PLC drives six circulating-air cooling fans in a six-chamber tunnel oven. Production passes on a conveyor belt whose instantaneous speed is measured in metres per minute. The control specification has two regimes and one retention requirement:

1. Low-speed protection mode (conveyor ≤ 10 m/min): Each of the six fans must run at a fixed 30 % reference. Operator HMI entries are locked out because, at low belt speed, the active product inside the oven can be thermally damaged if the cooling airflow drops below the protection value.
2. Normal operation mode (conveyor > 10 m/min): Each fan follows its own operator-entered HMI setpoint (0–100 %), used for fine-tuning individual chamber airflow for product-quality reasons.
3. Retention requirement: When the conveyor stops or drops below the threshold, the latest operator-entered values must be retained so that, when the conveyor accelerates through 10 m/min again, those exact values are restored to the live setpoints. Operators expect the line to "come back exactly where they left it".

Prerequisites

Hardware

• SIMATIC S7-300 CPU with retentive memory. Candidates are CPU 314C-2 PN/DP (integrated AI/AO, cost-effective), CPU 315-2 PN/DP (larger work memory, PROFINET), or CPU 317-2 PN/DP for high-end applications. Refer to the SIMATIC S7-300 Automation System, Hardware and Installation manual for order numbers, slot rules, and wiring practices.
• SM 331 analogue input for the conveyor speed transmitter (4–20 mA, two-wire). Typical order number: 6ES7331-7KF02-0AB0 (8 AI, 12-bit, suitable for slow process signals).
• SM 332 analogue outputs for the six fan drives. Two SM 332 AO 4 modules (6ES7332-5HD01-0AB0) cover six channels; the integrated AO of a CPU 314C-2 PN/DP provides the seventh channel.
• Operator panel: SIMATIC HMI such as KTP1200 Basic, TP700 Comfort, or MP 377. Configure with WinCC flexible 2008 SP5 (classic S7-300) or TIA Portal V18 with the S7-300 add-on package.
• Speed sensor: incremental encoder + FM 350-1 counter, or a 4–20 mA tachometer scaled 0–50 m/min. Use shielded, twisted-pair cable with shield grounded at the cabinet entry only.
• Six VFDs (e.g., SINAMICS V20 or G120) configured for analogue setpoint with 4–20 mA input and "fallback to last value" behaviour on signal loss.

Software

• STEP 7 V5.6 + SP2 (classic project) or TIA Portal V18 with the S7-300 HSP.
• Programming languages: STL (Statement List) shown for the kernel; LAD (Ladder) shown for the multiplexer; SCL is available on CPU 314C-2 and higher.
• WinCC flexible 2008 SP5 or TIA Portal WinCC for HMI screens.

System Architecture and I/O Assignment

The signal list below is a starting point; the engineer must reconcile it with the actual SM 332 channel assignments on the rail.

State-Machine View

Model the system as two states and one snapshot memory. Drawing this on paper before coding prevents most of the back-and-forth in the original engineering discussion.

Conveyor Speed Scaling (FC1)

The 4–20 mA analogue input is linear 0–50 m/min. Convert raw to engineering units with the standard Siemens norm scaling. The function below is one-shot in OB1 and easy to retune by changing the two constants.

// FC1 - Scale AI 0..50 m/min
//   Input : IW0  (0..27648 Siemens norm)
//   Output: MD100 REAL m/min
L   IW0
ITD
DTR
L   5.000000e+001
*R
L   2.764800e+004
/R
T   MD100

The 10 m/min threshold therefore corresponds to raw 10 × 27648 / 50 = 5529.6, rounded to 5530. Using REAL scaling throughout avoids integer rounding and gives clean hysteresis boundaries.

Hysteresis Parameters

STL Implementation (FC20 - Hysteresis and Debounce)

The compact STL kernel suggested in the original exchange handled the multiplexer only. Production code must add hysteresis, debounce, and snapshot. The complete FC20 follows.

// FC20 - Mode flag with hysteresis and stop debounce
//   Input : MD100   rConvSpeed  (REAL m/min)
//   Output: M10.0   bOperatorMode
//           M10.1   bConvStopped (latched after tStopDbnc)
A   M 10.0          // currently in operator mode?
JC  MODE_OP         // yes - test only high threshold
// currently in protection mode (30 %)
L   MD 100
L   1.050000e+001   // rSpdOn = 10.5
>R
=   M 10.0          // raise operator mode
JU   END

MODE_OP:
L   MD 100
L   9.500000e+000   // rSpdOff = 9.5
<R
=   M 10.0          // drop operator mode if below low threshold
END: NOP 0

// Conveyor-stop debounce 2 s
A   M 10.0          // speed is below threshold
O   M 10.1          // or already latched
AN   T 1
L   S5T#2S
SD   T 1
A   T 1
=   M 10.1          // bConvStopped - used by FC30 retention

The 5-line STL from the original engineering exchange, reproduced verbatim for traceability:

L   iConvSpeed
L   10
>I
L   iOperatorManipulatedSetpoint
JC  _001
L   iFixedSetpoint
_001: T  iSetpointFanSpeed

This compact snippet is the multiplexer only; in a real S7-300 oven control, the snapshot (FC30) and hysteresis (FC20 above) are mandatory to satisfy the operator's requirement that the latest values be stored when the belt stops and restored when it moves again.

STL Implementation (FC30 - Snapshot / Restore)

FC30 keeps two retentive data blocks: DB2 holds the live HMI values, DB3 holds the snapshot at the moment of the most recent conveyor stop. On a transition from PROTECTION to OPERATOR, the snapshot is copied back to DB2 so that operator setpoints resume exactly where they were.

// FC30 - Capture operator setpoints on belt stop, restore on belt start
// Trigger 1: M10.1 rising edge  (stop latched)  -> DB2 -> DB3
// Trigger 2: M10.0 rising edge  (operator mode) -> DB3 -> DB2
A   M 10.1
FP  M 11.0         // one-shot on stop rising
JCN NO_CAP
L   DB2.DBD0   T DB3.DBD0     // fan 1
L   DB2.DBD4   T DB3.DBD4     // fan 2
L   DB2.DBD8   T DB3.DBD8     // fan 3
L   DB2.DBD12  T DB3.DBD12    // fan 4
L   DB2.DBD16  T DB3.DBD16    // fan 5
L   DB2.DBD20  T DB3.DBD20    // fan 6
NO_CAP: NOP 0

A   M 10.0
FP  M 11.1         // one-shot on operator-mode rising
JCN NO_RES
L   DB3.DBD0   T DB2.DBD0
L   DB3.DBD4   T DB2.DBD4
L   DB3.DBD8   T DB2.DBD8
L   DB3.DBD12  T DB2.DBD12
L   DB3.DBD16  T DB2.DBD16
L   DB3.DBD20  T DB2.DBD20
NO_RES: NOP 0

STL Implementation (FC10 - Six-Channel Multiplexer)

// FC10 - Active setpoint generator (six fans)
//   M10.0 = 1 -> use DB2 (operator) setpoints
//   M10.0 = 0 -> use fixed 30 % protection value
//   Outputs: MD110, MD114, MD118, MD122, MD126, MD130
A   M 10.0
JCN PROT
L   DB2.DBD0    T MD110
L   DB2.DBD4    T MD114
L   DB2.DBD8    T MD118
L   DB2.DBD12   T MD122
L   DB2.DBD16   T MD126
L   DB2.DBD20   T MD130
JU   END

PROT:
L   3.000000e+001
T   MD110
T   MD114
T   MD118
T   MD122
T   MD126
T   MD130
END: NOP 0

OB1 then scales the REAL percent (0–100) to the analogue-output raw value (0–27648) and writes QW0, QW2, QW4, QW6, QW8, QW10. Standard scaling: raw = percent × 276.48. Clip the result to [0, 27648] before assigning to the PQW to prevent signed-overflow warnings on the SM 332.

Ladder (LAD) Alternative

For teams that prefer LAD, the same logic fits in five networks of OB1. A compact version of the multiplexer:

Network 5 - Fan 1 setpoint
|----[ M10.0 ]-----------------( MOV   DB2.DBD0 -> MD110 )----|
|----[ /M10.0 ]----------------( MOV   30.0    -> MD110 )----|

Repeat the rung five more times, replacing DB2.DBD0 with DB2.DBD4 / 8 / 12 / 16 / 20 and MD110 with MD114 / 118 / 122 / 126 / 130. Use a separate network with a comparator (GE 10.5, LE 9.5) and SR flip-flop to implement the hysteresis in pure LAD if STL is not allowed by the site standard.

SCL Variant (CPU 314C-2 and higher)

// FC20 in SCL
IF bOperatorMode THEN
    IF rConvSpeed < rSpdOff THEN bOperatorMode := FALSE; END_IF;
ELSE
    IF rConvSpeed > rSpdOn THEN bOperatorMode := TRUE;  END_IF;
END_IF;
bConvStopped := bConvStopped AND (rConvSpeed < rSpdOff);
IF rConvSpeed < rSpdOff THEN
    tStop(IN:=TRUE, PT:=T#2S);
ELSE
    tStop(IN:=FALSE);
END_IF;
bConvStopped := tStop.Q OR bConvStopped;

SCL reads more like the engineering specification and is preferable on newer S7-300 CPUs and on S7-1200/1500 migrations.

HMI Integration

• Six I/O fields, each tagged to DB2.DBD0..DBD20. Set limits 0.00–100.00 % with one decimal place, behaviour "Output"/"Input/Output".
• Disable input on each field when M10.0 = 0 (operator mode off). Use the HMI "Enable" property bound to M10.0. This satisfies the requirement that "operator does not change these values because product inside oven can be damaged".
• Status indicator: "Conveyor speed = X.X m/min" bound to MD100; a green lamp "Operator mode active" bound to M10.0; an amber lamp "Belt stopped" bound to M10.1.
• Recipe view: each recipe row contains the six fan setpoints. Persist recipes on the panel's flash; copy to DB2 on "Load recipe". Keep DB3 as the snapshot, untouched by recipe load.
• Alarm view: alarm "Conveyor speed out of range" when MD100 > 55 or MD100 < 0 (transmitter fault).

Commissioning and Verification

1. Wire check: with the CPU in STOP, force IW0 = 0; verify the transmitter wiring per the SM 331 wiring diagram in the S7-300 hardware manual. Confirm 24 V is present at the two-wire loop terminals and that the shield is grounded at the cabinet entry.
2. Scale check: drive the transmitter with a calibrator at 4 mA and 20 mA; confirm MD100 reads 0.0 and 50.0 ± 0.1 m/min in the VAT table. If the reading is inverted, check whether the SM 331 channel is configured for 4–20 mA or 0–20 mA in HW Config.
3. Threshold step test: with MD100 = 10.5, the mode flag M10.0 must set; reduce to 9.5, the mode flag must reset. Verify in VAT with online monitor. Walk the speed slowly through the band (0.1 m/min steps) and confirm no chatter.
4. Snapshot test: in RUN, change DB2.DBD0 from 50 % to 75 %; pull MD100 to 0 for more than 2 s; observe DB3.DBD0 = 75 %; raise MD100 back to 12; observe DB2.DBD0 = 75 % after restore.
5. Fan sanity: with the VFDs in manual, command the PLC outputs from VAT; confirm each PQW follows the setpoint with the expected 0–10 V or 4–20 mA scaling.
6. HMI lockout test: drop speed below 9.5; confirm operator I/O fields go grey and cannot be edited. Click into a field and try to type — WinCC should reject the input.
7. Power-cycle retention test: enter six setpoints, run CPU STOP/RUN several times; confirm values remain unchanged thanks to the retentive DBs.
8. End-to-end test: run the conveyor at 5 m/min (all fans 30 %), accelerate to 12 m/min (all fans switch to operator values), stop the conveyor for 10 s (fans revert to 30 %, snapshot stored), restart (operator values restored).

Troubleshooting Matrix

Field-Proven Caveats

• Slipping encoder roller: if the encoder is mounted on a slipping pulley, the apparent speed may bounce between 9 and 11 m/min even though the belt is steady. Increase Δ to 2 m/min for such installations, or add a PT1 low-pass on MD100 with T = 500 ms.
• Soft-starter ramp: during VFD ramp-up the speed may dwell at exactly 10 m/min. The 2-second debounce absorbs this and keeps the fans in protection mode until the belt has clearly stabilised.
• Operator misuse: tying the HMI input enable directly to M10.0 prevents the operator from changing setpoints during low-speed protection, matching the requirement that "operator does not change these values because product inside oven can be damaged".
• Setpoint sanity: clamp HMI input to [5 %, 100 %] to avoid sending 0 % to a VFD that interprets it as "stop", which would trip the drive with F0001 (overcurrent on ramp-down) on many inverters.
• Three-segment thinking trap: the discussion in the original engineering thread went back and forth because the requirement is not a simple comparator; it is a state machine with two modes and a snapshot memory. Drawing the three boxes (current mode, snapshot, multiplexer) before coding saves hours.
• CPU OB1 priority: do not place FC20/FC30/FC10 in OB100 (startup) because the conveyor speed is not yet valid; OB1 cyclic execution is correct.
• Diagnostic interrupts: enable OB82 for analogue-channel diagnostics. Without it, a broken 4–20 mA loop will not raise an SF, and the conveyor speed will sit at 0 indefinitely, locking the system into protection mode.

CPU Selection Comparison

Variants and Platform Notes

• S7-1200 / S7-1500 (TIA Portal): The same logic translates directly to SCL with a structure IF rSpd > rSpdOn THEN bOp := TRUE; ELSIF rSpd < rSpdOff THEN bOp := FALSE; END_IF;. Replace M10.0 with a retentive global DB tag "Mode".bOperatorMode. Edge detection uses R_TRIG / F_TRIG IEC instances.
• LOGO! 8: Implementable with the analogue threshold trigger and a soft-retentive shift register, but six independent channels plus retention is at the upper limit of LOGO! memory.
• SINAMICS V/F fans: drive setpoint source should be "Analogue input" with the AI wired from the SM 332 output. Avoid command source "Fieldbus" because fieldbus fault defaults to 0 % on most inverters, killing the protection setpoint.
• PROFINET variant: if the VFDs are G120 with PROFINET, replace the SM 332 AO path with PROFINET telegram 1 (setpoint) and keep the protection value 30 % in the drive's local fallback parameter (P0822 in some variants). The PLC hysteresis logic remains unchanged.

Glossary

FAQ