How-to
S7-300 PID control for a modulating burner motor
Short answer
Run one FB41 CONT_C loop on the hottest hood's PT100, scale all five probes to °C, and turn the 0-100% output into open/close pulses that follow the M941B's 0-10 V pot feedback to a 2% deadband, through 24 VAC interposing relays. The gas management system keeps the burner start sequence and its permissives; the PID never touches them.
On this page
- Which loop structure replaces five Eurotherm controllers on one shared motor?
- How do you scale each PT100 to degrees Celsius in STEP 7?
- How do you set up FB41 CONT_C for a slow duct temperature loop?
- How does a 0-100% PID output drive a 24 VAC open/close actuator?
- What happens when a PT100 fails or the burner drops out?
- What do you test before film goes through the drier?
- How much of the loop belongs on the HMI?
- What do you commit to on this retrofit?
- Questions asked
You can retire the Eurotherm 818 loops and let the S7-300 modulate the burner motor, but only after you fix three boundaries first: one demand authority for the shared M941B actuator, five PT100 probes scaled to °C that the PID can actually use, and a pulse-and-feedback servo that turns the 0-100% controller output into 24 VAC open/close commands. The gas management system keeps doing what it does today — it interposes the burner start sequence on a relay fired from the PLC — and the PID never touches that path. The sections below give the loop structure, the FB41 CONT_C configuration, the actuator arithmetic with real numbers, and the commissioning order that proves each piece before film goes through the drier.
Which loop structure replaces five Eurotherm controllers on one shared motor? #
Use exactly one PID loop for the single M941B: make the process variable the highest of the five scaled hood temperatures, and let that one 0-100% demand drive the actuator, because five independent controller outputs wired to one motor would fight each other continuously.
One actuator accepts one position, so the strategy has to name a single authority before any block is written. Before you delete the Eurotherms, trace which controller commands the M941B today and whether the others trim it; whatever that authority is, reproduce it deliberately in the PLC instead of assuming the five loops acted independently. With the highest-hood PV, the loop regulates the worst-case hood, and setpoints for the other hoods become per-hood alarms rather than competing demands on a shared motor.
If a future rebuild gives each hood its own actuator, the structure changes to one FB41 instance per hood with its own relay pair, and only then. Until that hardware exists: one loop, one demand.
How do you scale each PT100 to degrees Celsius in STEP 7? #
Put each PT100 on a dedicated RTD channel configured in HW Config for the wiring you actually ran, then scale the channel's integer through FC105 into a REAL in °C, and prove each reading against a reference thermometer before any loop sees it.
RTD measurement belongs on an RTD-capable input, the same discipline that makes FX5U thermocouple readings need a dedicated TC input on a different controller family: the excitation and lead compensation live in the channel hardware, not in software. S7 analog inputs report on the convention where 27648 counts is 100% of the nominal range, and FC105 SCALE converts that integer into engineering units between LO_LIM and HI_LIM.
Worked example, with the channel ranged 0.0 to 250.0 °C and a raw value of 13824 counts: 13824 ÷ 27648 = 0.500, and 0.500 × (250.0 - 0.0) = 125.0 °C, so FC105's OUT must read 125.0 °C. Module open-circuit and short-circuit diagnostics flag a failed probe; pass that fault bit into the logic and never let a frozen value ride into the PID as a healthy PV.
How do you set up FB41 CONT_C for a slow duct temperature loop? #
Call FB41 CONT_C from a fixed-interval cyclic interrupt OB, set CYCLE to match that real interval — T#500MS if you call it every 500 ms — and feed PV_IN and SP_INT in °C so the gain, Ti and Td act on real degrees.
FB41 CONT_C is the continuous PID block in STEP 7's standard PID control package; parameterize it through the PID parameter assignment tool (classic STEP 7 menu path: Start > Simatic > Step7 > Assign PID Control parameters) or directly in the instance data block. CYCLE is the sample time the block uses to scale its integral and derivative math, so it must equal the actual call interval — the same reason Logix 5000 PID loops run in a periodic task instead of a variable-scan context.
Hood ducts are slow thermal masses, so start from GAIN := 1.0, TI := T#120S, TD := 0 with D_SEL off, and tune from this duct's response rather than another machine's numbers. Before enabling output, prove direction on the bench: with PV below SP, the actuator must move toward more heat on this burner train, and the action is set accordingly. Commissioning acceptance: hold SP within ±2 °C steady-state; if the film specification is tighter, tighten the acceptance figure, not the starting gains.
How does a 0-100% PID output drive a 24 VAC open/close actuator? #
Treat the FB41 output as a position demand in percent, read the M941B's 0-10 V pot as measured position, and energize the open or close command through interposing relays until the two agree within a 2% deadband — never both at once.
LMN in percent is a number, not a signal this actuator understands. The pot feedback lands on an analog input; FC105 with LO_LIM := 0.0 and HI_LIM := 100.0 turns it into percent directly. The servo logic compares demand with measured position, energizes one relay when the error exceeds the deadband, and drops both when it does not, with the mutual interlock in the logic so open and close cannot be true together.
Worked example: LMN = 62.0% and the feedback raw value is 15500 counts. Position = 15500 × 100.0 ÷ 27648 = 56.06%, which is 15500 ÷ 27648 × 10 V = 5.61 V. Error = 62.0 - 56.06 = 5.94%, above the 2.0% deadband, so energize the open relay and stop once measured position sits between 60.0% and 64.0%, i.e. 6.00 to 6.40 V at the input.
Jog the actuator to each end and scale the 0% and 100% points from the measured endpoint voltages, because pot actuators rarely span exactly 0.00 to 10.00 V; take travel time from the actuator's documentation and confirm it on the machine, since it sets how long a pulse runs before the next correction makes sense. The Eurotherm supplied the motor 24 VAC today, so reuse that supply: PLC output points switch interposing relays with 24 VAC coils whose contacts carry the actuator circuits, because the output card is not a 24 VAC source itself. If a retrofit later swaps in an actuator that takes a continuous positioning signal, which modulating valve actuator takes 0-10 V in and 4-20 mA out explains how LMN then maps straight onto the input and the pulse servo disappears.
What happens when a PT100 fails or the burner drops out? #
The temperature PID modulates only: when a PT100 fails or its reading goes implausible, switch FB41 to manual, hold the actuator where it was, raise an alarm, and leave every start permissive and shutdown on the gas management system's side of the relay.
Probe fault: set MAN_ON true, freeze MAN at the last position, and alarm on the HMI, because a demand computed from a dead probe is worse than no demand. Burner enable dropped: park the position demand at the burner's minimum firing position — a figure from the burner documentation — and keep the loop in manual until the gas system re-enables. Reject PV values outside the calibrated hood range before they reach PV_IN.
Safety note: never connect the PID output, the pulse servo, or any HMI tag to the gas train's start permissives, flame safeguard, or shutdown path; the burner management system holds sole authority there, and the PLC's involvement ends at the one relay into its start sequence.
What do you test before film goes through the drier? #
Prove the chain in manual before you close it: calibrate both actuator endpoints, confirm direction, inject a probe fault, and run the burner start sequence with the PID output inhibited — then and only then hand the loop automatic control.
- With the burner inhibited, jog the actuator fully open and fully closed and enter the two measured feedback voltages as the 0% and 100% endpoints.
- Command mid-travel in manual and confirm the servo stops inside the deadband with both command relays off.
- With PV below SP, verify the demand moves the actuator toward more heat; correct the action before anything else.
- Disconnect one PT100 with the process in a safe state and confirm the fault bit, the manual hold, and the HMI alarm.
- Fire the burner start sequence three times through the gas management system with the PID output inhibited, proving the PLC relay interposes it and cannot bypass it.
- Enable automatic under observation, apply a small setpoint step, and trend PV, LMN, and measured position until the loop holds setpoint steady inside the acceptance band from the FB41 section.
No automatic operation until every step passes; a step that fails sends you back to the section above that owns it.
How much of the loop belongs on the HMI? #
Give the HMI exactly three jobs on this loop: write SP_INT in °C, switch between automatic and manual through MAN_ON, and display the five hood temperatures with their fault bits — and protect those tags from stray writes.
When an operator takes manual, write the current measured position into MAN at the same instant so the actuator does not jump; on the way back to automatic, initialize through the block's start-value inputs (I_ITL_ON with I_ITV at the measured position) so the output does not step. Setpoints in °C match PV_IN units exactly, with no conversion at the screen. Interlock or privilege the mode and setpoint writes with the same discipline as safe Modbus writes to PLC output registers from SCADA, because a stray write to MAN_ON or SP_INT moves a live burner loop. Screen layout itself belongs to the HMI library; the controls side ends at these tags.
What do you commit to on this retrofit? #
Commit to one FB41 loop on the hottest hood, pulse control through 24 VAC interposing relays against a measured 0-10 V feedback scale, and a hard boundary at the gas train relay — with one loop per hood only if each hood ever gets its own actuator.
- Five hoods, one M941B: one FB41 instance, PV is the highest of the five scaled hood temperatures, SP set for the governing hood.
- Per-hood actuators in a future rebuild: one FB41 instance per hood, one relay pair each, no shared demand.
- Continuous-input actuator in a swap: LMN scaled straight onto the positioning input and the pulse servo deleted.
- Probe fault: manual, hold position, alarm. Burner enable dropped: park at minimum firing position until the gas system re-enables.
- Starting gains and acceptance exactly as set out in the FB41 section, tuned on the duct.
Every number above is a starting point you prove on the machine; the structure — one demand authority, scaled inputs, position servo, sealed gas boundary — is the part that does not change.
Questions asked
What happens if FB41's CYCLE does not match the real call interval?
CYCLE is the sample time FB41 uses in its integration and derivative math. If you set it to T#1S but actually call the block every 500 ms, integral action runs at twice the rate the Ti value suggests, so the loop overshoots as though Ti were halved. Set CYCLE from the actual cyclic interrupt OB interval, and call the block there, not in OB1 where scan time varies.
How many FB41 instances does this drier need?
One per controlled variable, each with its own instance data block. This design uses one instance for the master hood loop. If you later fit a dedicated actuator per hood, create one instance DB per hood with its own scaled PV and setpoint, and never route two instances' outputs to the same command relay pair, because the pulse servo assumes it owns its actuator.
Why does the actuator hunt around setpoint and how do you stop it?
Hunting means the deadband is smaller than the position error the mechanics create. Measure backlash by jogging the actuator open until the damper crank just moves, then closed until it moves back, and read the feedback voltage difference between the two; set the deadband just above that. If it still hunts, add a minimum off-time between pulses so the motor stops fully before the next correction.
Do you still need FC105 if the RTD channel is already in a temperature range?
No. When the analog channel is configured for a temperature measuring range, the module data manual defines the integer-to-°C conversion, and the value converts straight to REAL or through FB41's own PV_FAC and PV_OFF inputs. FC105 with LO_LIM and HI_LIM is the cleaner path when scaling from a normalized range or from a plain 0-10 V input such as the pot feedback.