Web Tension Control Explained: Methods, Sensors, and PLC Control

Web tension control is the practice of maintaining a precise, consistent pulling force on a continuous strip of material — the web — as it moves through a converting, printing, or coating machine. Get it wrong and the web wrinkles, stretches, breaks, or goes off-register. Get it right and every meter of film, foil, paper, or fabric comes off the line within specification.

This guide covers the engineering fundamentals: what tension is and why it matters, the two control architectures, how dancer arms and load cells feed back tension data to a PLC, the three distinct zones every line has, taper tension on winders, and the drive coordination strategies that hold it all together.

Table of Contents

1. What Web Tension Control Is
2. Why Tension Matters in Converting and Printing
3. Open-Loop vs Closed-Loop Tension Control
4. Feedback Methods: Dancer Arm vs Load Cell
5. The Three Tension Zones
6. Taper Tension on Winders
7. Torque vs Speed Control Strategies
8. The Controls View: PLC and Drive Implementation
9. Frequently Asked Questions

What Web Tension Control Is

A web is any continuous flexible material — paper, plastic film, foil, nonwoven fabric, wire, or textiles — fed from a roll, processed across one or more driven nip rolls or idlers, and rewound or cut at the exit. Web tension is the force, measured in Newtons or pounds-force, applied lengthwise through the web between two driven sections.

Tension arises whenever one driven section runs at a slightly different speed than the adjacent section. Increase the speed differential and tension rises. Decrease it and the web goes slack. Web tension control is the feedback and actuation system that continuously corrects that differential — or the torque producing it — to keep tension at a target value.

Key parameters in any tension control system:

• Setpoint — the target tension, typically in N or lbf, sometimes expressed in N/m (tension per unit width)
• Process variable — the measured tension from a load cell or inferred position from a dancer
• Manipulated variable — drive speed (speed control) or drive torque (torque control)
• Web geometry — width, thickness, modulus, and roll diameter all affect how tension behaves

Tension control sits at the intersection of mechanical engineering, drive technology, and PLC programming. The PLC handles the control logic; the drive executes torque or speed commands; the mechanical design — roll alignment, wrap angle, surface finish — determines how much the control system has to work.

Why Tension Matters in Converting and Printing

Every material has an elastic limit. Run a web above it and you permanently deform the material, causing gauge variation downstream. Run it too loose and registration marks drift, coating weight varies, and the web sags onto guides causing scratches.

Industries where tension errors are costly:

Industry	Effect of high tension	Effect of low tension
Flexible packaging film	Neck-down, gauge loss, edge tear	Wrinkle, blocking, registration error
Paper printing	Web break, dot gain variation	Slack side, head crash on inkjet
Aluminum foil	Elongation, gauge variation	Fold, crease, winding defect
Lithium-ion battery electrodes	Crack in active coating	Delamination, air entrapment
Nonwoven fabric	Fiber breakage, width loss	Puckers, fold-over at nips

In precision slitting, a tension error of a few percent can push roll hardness outside specification. In gravure printing, uncontrolled tension shift causes color-to-color misregister measured in fractions of a millimeter. This is why tension control is a first-order process variable, not an afterthought.

Open-Loop vs Closed-Loop Tension Control

Open-Loop Tension Control

In open-loop (also called feedforward) tension control, the PLC or drive calculates the required torque or speed ratio from known machine parameters — roll diameter, line speed, material width — without measuring actual tension.

How it works: The controller estimates the roll diameter in real time (by integrating material thickness against roll revolutions, or from an ultrasonic sensor), then computes the torque needed to produce the target tension:

Torque = Tension_setpoint × Radius

The drive receives a torque command proportional to the current radius. No sensor measures the result.

Advantages: Simple, no sensor to calibrate, fast response to speed changes, immune to sensor noise.

Disadvantages: Accumulates error from diameter estimation drift, cannot compensate for eccentricity, material variation, or mechanical losses. Accuracy is typically ±10–20% without additional correction.

Open-loop control is used on unwind stands with well-characterized materials and on low-value applications where occasional wrinkle is acceptable.

Closed-Loop Tension Control

Closed-loop control measures tension directly (load cell) or indirectly (dancer position) and feeds the error back through a PID controller to correct the drive command.

The PLC PID tuning loop continuously computes:

Output = Kp × error + Ki × ∫error dt + Kd × d(error)/dt

Where error = setpoint − measured_tension. The output drives either the speed reference or torque reference of a variable frequency drive.

Advantages: Compensates for diameter variation, material property changes, eccentricity, and mechanical losses. Typical accuracy ±1–3% of setpoint.

Disadvantages: Requires a sensor (cost, maintenance), proper PID tuning, and careful loop design to avoid instability — particularly on highly elastic webs where the process gain changes with speed.

Most production converting lines use closed-loop control on the unwind and rewind, with speed-ratio (draw) control across the process section.

Feedback Methods: Dancer Arm vs Load Cell

The two dominant tension feedback devices have different operating principles, and choosing between them shapes the entire control architecture.

Dancer Arm

A dancer is a roll mounted on a pivoting arm or linear carriage. The web wraps over the dancer roll. When tension rises, the web pulls the dancer toward its high-tension limit; when tension falls, the dancer moves toward its low-tension limit.

The dancer position is measured by a potentiometer, linear encoder, or analog angle transducer and fed to the PLC as a 0–10 V or 4–20 mA signal.

Control logic: The PLC controls the upstream or downstream drive to keep the dancer at a mid-position setpoint. The dancer acts as a mechanical integrator — it absorbs momentary speed mismatches as position change rather than tension spikes. This makes dancer systems inherently stable and forgiving of drive speed ripple.

Best suited to:

• Low-tension webs (films under 50 N/m) where load cells are hard to calibrate
• Applications with frequent splices or speed changes
• Lines where the dancer mass can be tuned to the web modulus
• Unwind and rewind zones

Limitations: The dancer adds inertia; its mass must be balanced against web tension. Heavy dancers on light webs cause tension oscillation. Stroke length limits the buffer capacity.

Load Cell

A load cell converts the force the web exerts on a sensing roll into an electrical signal. A bridge-mounted roll rests on two strain gauge load cells; the web wraps partially around it. The measured force, corrected for wrap angle and roll weight, equals web tension.

Load cell output is typically a millivolt signal amplified to 4–20 mA or ±10 V for the PLC analog input.

Control logic: The PLC reads tension directly and runs a PID loop against the tension setpoint. There is no mechanical buffer — every tension transient reaches the controller immediately.

Best suited to:

• High-speed precision lines (printing, coating, battery manufacturing)
• Applications requiring direct tension measurement in N or lbf
• Process sections where dancer geometry is impractical
• Wide webs where dancer inertia would be prohibitive

Limitations: Sensitive to roll contamination, bearing friction, and vibration. Requires periodic calibration. PID tuning is more demanding because there is no mechanical filtering.

Comparison Summary

Attribute	Dancer	Load Cell
What is measured	Dancer position (indirect)	Web force (direct)
Mechanical filtering	Yes — dancer absorbs spikes	No — direct coupling
Calibration requirement	Low	Regular (monthly typical)
Typical tension range	1–500 N	5–50,000 N
PID tuning difficulty	Moderate	Higher
Response to splices	Excellent	Requires feedforward
Cost	Lower	Higher

In practice, many lines combine both: a dancer on the unwind for splice buffering and load cells in the process section for precision measurement.

The Three Tension Zones

Every web handling line, regardless of industry, has three fundamentally different tension zones. Understanding them is the foundation for drive coordination and control architecture.

Zone 1: Unwind

The unwind pays off material from a parent roll. The drive must apply a braking torque — or a regenerative retarding force — to hold back the accelerating web. As the roll unwinds and the diameter decreases, the same line speed requires the roll to rotate faster, and the same tension requires less torque.

Control objective: Maintain constant unwind tension while accommodating the continuously decreasing roll diameter.

Typical control method: Open-loop torque control with diameter calculation, or closed-loop with a dancer that buffers splice events.

Key challenge: The inertia of a full parent roll can be enormous. During acceleration and deceleration, feedforward inertia compensation is required to prevent tension spikes.

Zone 2: Process Section

The process section spans everything between the unwind and the rewind — coating stations, printing decks, lamination nips, ovens, slitters. Material is processed at constant or variable speed. Tension in this zone affects registration, coating weight, and product geometry.

Control objective: Maintain a stable, known tension through each process step. Tension may be intentionally different between subsections (e.g., lower in the oven to allow thermal expansion).

Typical control method: Speed-ratio (draw) control between driven nip rolls, with load cells providing trim correction. The process section is essentially a cascade of speed-locked zones, each with a small trim capability.

Key challenge: Oven sections stretch the web thermally. Lamination nips add material. The control system must account for these sources of length change.

Zone 3: Rewind

The rewind winds finished material onto a core. As the roll builds in diameter, the same line speed requires the roll to rotate more slowly, and the same tension requires more torque.

Control objective: Maintain consistent rewind tension while accommodating increasing diameter, and optionally apply taper tension to control roll hardness.

Typical control method: Closed-loop torque control with diameter calculation, often with a surface-winding or center-winding drive, plus load cell or dancer feedback.

Key challenge: Roll hardness — how tightly the material is wound — depends on both the tension at the nip and the nip load if a lay-on roll is present. Winding too tight traps air and causes core crushing; winding too loose causes telescoping.

Taper Tension on Winders

A taper tension profile reduces the tension setpoint progressively as the rewind roll diameter increases. It compensates for the increasing number of wound layers, which would otherwise clamp inner layers so tightly they deform or block.

The standard taper tension calculation is:

T_current = T_core × (D_core / D_current) ^ taper_factor

Where:

• T_core — the tension setpoint at the bare core
• D_core — the bare core diameter
• D_current — the current roll diameter (calculated or measured)
• taper_factor — typically 0.0 to 1.0; 0 = no taper, 1 = full inverse-diameter taper

A taper factor of 0.5 is a common starting point for film rewinding. The optimal value depends on material modulus, roll width, and winding speed.

Implementing taper tension in a PLC: The PLC calculates D_current from the length wound (if material thickness is known) or from an ultrasonic or laser diameter sensor. The tension setpoint register is updated each scan cycle using the formula above. The torque command to the drive is:

Torque_cmd = T_current × (D_current / 2) / (mechanical_efficiency)

Taper tension is a form of open-loop compensation layered onto a closed-loop tension controller — the base tension loop still runs, but its setpoint changes with diameter.

Torque vs Speed Control Strategies

Web tension can be regulated by controlling torque or by controlling speed. Understanding when to use each strategy is fundamental to drive selection and PLC programming.

Torque Control

The drive is commanded to produce a specific shaft torque regardless of speed. Tension is directly proportional to torque and inversely proportional to radius:

T = Torque / R

Torque control is natural for unwinds and rewinds where the roll radius is continuously changing. The PLC computes the required torque from the desired tension and the current diameter estimate, then sends a torque reference to the drive.

For VFD programming and PLC-drive integration, most modern drives accept a torque reference over fieldbus (PROFINET, EtherNet/IP, EtherCAT) or via an analog 4–20 mA input. The drive's speed regulator is disabled or replaced by a speed-limit clamp to prevent runaway in the event of web break.

Speed Control with Tension Trim

In the process section, drives run in speed control, locked to a line speed master reference. A small tension trim — typically ±5% of line speed — is added from the PID tension controller to maintain the desired draw between consecutive sections.

Draw is the deliberate speed differential expressed as a percentage:

Draw (%) = (Section_N_speed - Section_(N-1)_speed) / Section_(N-1)_speed × 100

A positive draw stretches the web slightly; a negative draw (let-off) relaxes it. Draw values of 0.1–2% are common in printing and coating.

Cascade control is used when load cells feed a tension PID whose output trims the speed reference:

Speed_ref = Master_speed × (1 + Draw_bias + PID_tension_trim)

Choosing Between Torque and Speed Control

Condition	Preferred strategy
Unwind / rewind with large diameter change	Torque control
Process section between nip rolls	Speed control with tension trim
Highly elastic web (rubber, stretch film)	Torque control, tight speed limits
Inelastic web (paper, foil)	Speed control is viable with load cell trim
Web break recovery	Speed control — torque control spools to max speed

The Controls View: PLC and Drive Implementation

Analog Input Scaling for Load Cells

A load cell amplifier typically outputs 4–20 mA. The PLC analog input module converts this to a raw integer (e.g., 0–27,648 on Siemens, 0–32,767 on Allen-Bradley). Scaling to engineering units:

Tension_N = (Raw - Raw_min) / (Raw_max - Raw_min) × (Force_max - Force_min) + Force_min

Zero and span calibration must account for the tare weight of the sensing roll. Most PLC platforms provide a dedicated scaling function block (SCALE in Studio 5000, FC105 in TIA Portal STEP 7, or SEL/NORM in TIA Portal V16+).

The load cell signal chain — strain gauge bridge → amplifier → analog input → PLC — introduces measurement delay. For high-speed lines above 400 m/min, using a drive with onboard tension control closes the loop at the drive's internal scan rate (typically 250 µs) rather than waiting for the PLC scan cycle (typically 10–50 ms).

Dancer Position Feedback

A dancer potentiometer outputs 0–10 V proportional to angular position. The PLC reads this as a position percentage (0–100%) and runs a position PID to keep the dancer at 50%.

Anti-windup is critical for dancer loops during machine stops: when the line decelerates to zero and the dancer hits its mechanical stop, the integrator saturates. On restart, the accumulated integral term commands a large speed step that either breaks the web or slams the dancer to the opposite stop. Clamp the integrator output to ±10% of line speed and reset it to zero on stop.

Diameter Calculation

The PLC estimates roll diameter from the ratio of line speed to roll surface speed:

D_current = (Line_speed_mpm / Roll_RPM) / π

Alternatively, if material thickness is known:

D_current = sqrt(D_core² + (4 × Length_wound × Thickness / π))

The first method requires accurate speed feedback from a line speed encoder or a reference section. The second method accumulates error if thickness varies. Production lines often use both and cross-check them; a divergence beyond a threshold flags a measurement fault.

Coordinating Multiple Drives

A web line with six driven sections requires coordinated acceleration and deceleration to maintain tension ratios during speed changes. The standard architecture is a line speed master — a virtual axis or a designated master drive — with all section drives following as a ratio of the master:

Section_N_speed = Master_speed × Gear_ratio_N × (1 + Trim_N)

Trim_N is the output of each section's tension or dancer PID, bounded to ±5%.

During acceleration, inertia compensation feedforward adds a torque term proportional to d(Master_speed)/dt to each unwind and rewind drive. Without it, the tension spikes during ramp-up and sags during ramp-down, often triggering web break faults.

For motion control architecture on multi-axis web lines, electronic line shafting — where all drives synchronize via a real-time fieldbus — has largely replaced mechanical lineshafts. EtherCAT or PROFINET IRT provides sub-millisecond synchronization, which is sufficient for lines running below 600 m/min.

Typical drive coordination fault ladder:

1. Monitor dancer position limits — if dancer reaches travel limit, trip the section drive
2. Monitor load cell reading against min/max tension limits — if outside window for >200 ms, trigger controlled stop
3. Monitor diameter calculation for step change (web break signature) — if diameter jumps >10% in one scan, trigger emergency stop
4. Synchronize all drive stop ramps to the same deceleration time to prevent tension collapse

Web tension control appears across process industries. Operators familiar with injection molding machine control and palletizer drive coordination will recognize similar PID and drive synchronization patterns applied in a linear, continuous-process context.

---

Frequently Asked Questions

What is web tension control?

Web tension control is the automated regulation of the pulling force (tension) on a continuous strip of material — a web — as it moves through a converting, printing, coating, or winding machine. A sensor (load cell or dancer) measures or infers the tension; a PLC or drive runs a control algorithm to keep that tension at a target value by adjusting drive speed or torque.

What is the difference between a dancer and a load cell?

A dancer is a movable roll that absorbs tension variations as position change — it measures tension indirectly by tracking its own displacement. A load cell is a strain-gauge-based force sensor that measures web tension directly in Newtons. Dancers provide inherent mechanical filtering and suit low-tension or high-splice-rate applications. Load cells provide direct, accurate tension measurement and suit high-speed precision lines. Many machines use both: a dancer on the unwind and load cells in the process section.

What is taper tension?

Taper tension is a programmed reduction in the winding tension setpoint as the rewind roll grows in diameter. It prevents inner wound layers from being crushed by the compressive stress of outer layers. The tension is typically reduced as an inverse function of roll diameter, with a taper factor (0–1) controlling the rate of reduction. A taper factor of 0.5 is a common starting point for film rewinding.

How does a PLC control web tension?

A PLC controls web tension by running a PID feedback loop. The process variable comes from a load cell (analog input scaled to Newtons) or a dancer position transducer. The PID output either trims the speed reference of a VFD (speed control with tension trim) or directly commands a torque reference to the drive (torque control). In multi-zone lines, the PLC also calculates roll diameter in real time to adjust torque commands as the roll builds or depletes, and coordinates all section drives against a single line speed master reference to maintain tension ratios during acceleration and deceleration.