How to Choose a Web Tension Control System

How to choose a web tension control system rarely fails at the component level. It fails because the machine was zoned wrong, or the money went to the wrong zone, or nobody wrote down what "good" means before the purchase order went out. Most buying guides start from a catalogue of controllers and sensors. This one starts from the web path, because your line geometry decides everything else: how many zones exist, which can be measured, and what each one is worth spending on.

The six steps below apply to new machines and retrofits on slitters, rewinders, presses, laminators and coaters. For the building blocks this article assumes, see What Is a Web Tension Control System?.

Step 1: Zone the Web Path Before You Buy Anything

A tension zone is a run of web between two controlling points. A zone has exactly one tension value, because there is nothing inside it to create a second one. That single fact sets your entire budget.

Count the tensions your process needs, not the rollers you happen to have. A coater wanting low tension through the dryer and higher tension at the laminating nip needs two zones, which means one isolating boundary between them — usually a driven nip or an S-wrap running slightly faster than the section before it. Each added zone costs another torque source, another control loop and quite possibly another drive.

Sketch the web path with a value above each span; merge adjacent spans that share a value. That sketch, not a controller datasheet, is what you should discuss with suppliers. The rule that every zone needs one speed master and one torque source is covered here.

Step 2: Pick a Topology Per Zone

Once each zone has a target tension and a known tolerance, the topology follows. Choose from what the zone must deliver rather than from what the machine already has:

If the priority is…	Topology	What you accept
Stable coarse tension on paper, board, nonwoven	Unwind brake or rewind clutch, open loop with roll-diameter tracking	Tension drifts with wear and humidity, typically 10–15%
Substrate takes permanent damage from tension error	Closed loop: load cell, controller, magnetic powder brake or magnetic powder clutch	Higher cost, calibration schedule, room for a measuring roller
Web changes length between spans (oven, UV, coating)	Dancer accumulator as both sensor and buffer	Inertia, added web length, careful arm geometry
Different tension either side of a process step	Split into two zones with a driven isolation nip, upper zone on draw control	One extra drive axis and one extra loop to tune

The first row is genuinely sufficient and far cheaper; Open Loop vs Closed Loop Tension Control covers the break-even. Row two must be specified as one chain — automatic web tension controller, pillow block tension sensor, cantilever tension load cell or through-shaft web tension sensor, plus the actuator. Mismatched ranges and signal levels are the usual reason a new loop underperforms.

Step 3: New Line or Retrofit — Two Different Orders

New machines and retrofits use opposite sequences, and mixing them up is expensive.

On a new line the order is process-first: what tension does the material need → zone map → topology per zone → size the duty → choose hardware → write acceptance numbers.

On a retrofit the order is feasibility-first, because the mechanical answer may be no. You can always add electronics; you often cannot add a sensor, because converting a roller into a measuring roller requires a free span, bearings you can replace, and a cable run that does not become an antenna.

Survey these four things before quoting anything:

1. A measurable roller. Is there an idler that can carry a load cell with a stable wrap angle either side?
2. Space for the actuator. Can the unwind shaft take a mounted brake, or does it need a coupling and bearing pillow?
3. A spare drive axis. A new zone boundary means a new motor; without one the cheap second zone triples in price.
4. Cable runs. Sensor distance to the controller, and whether loop wiring can be separated from drive cabling.

A retrofit usually keeps the working open-loop torque device and adds measurement. Packages such as the XW micro tension controller TC200 and the compact tension controller exist because the constraint is panel space and wiring, not control performance.

Step 4: Spend the Sensor Budget Where Error Becomes Scrap

The common budgeting mistake is buying a load cell for every zone. Put feedback only where a tension error becomes rejected product in that zone. Score each zone 0 to 2:

• Does the material take permanent deformation here? Film stretches, coatings crush, foil breaks.
• Is this zone immediately upstream of a quality-critical step — print register, slit width, coating weight?
• What does one minute of off-spec output cost at this point?

A score of 4–6 means closed loop is justified. Two or three means open loop with diameter compensation. Zero or one means leave it manual and spend elsewhere. On a typical slitter-rewinder this usually puts feedback on the unwind-to-slitting zone and the rewind zone, and nothing in between.

Two rules save money. Zones that never run different recipes at once can share one multi-channel controller. And since a second loop added later costs mostly mechanical work and downtime, pre-wire and pre-mount for it now even if you buy one zone today.

Step 5: Size Across the Roll, Not Just at the Setpoint

Every device must work at both ends of the roll. The actuator delivers T = F × R, so it must reach maximum tension at full radius and still resolve minimum tension on the core. Turndown is roughly (F_max × R_full) ÷ (F_min × R_core): a 76 mm core to 1000 mm is about 13:1 on radius alone, and a 2:1 taper profile pushes it near 26:1. Few actuators span that and still control cleanly at the bottom, which is why low-tension cores feel sloppy.

The binding constraint is usually the sensor, not the actuator. The smallest force you must resolve sits at minimum tension and minimum radius, so size the sensor for that floor rather than the ceiling — lost resolution at the core is far more common than an overloaded cell.

Step 6: Write the Acceptance Numbers Into the Purchase Order

Nothing protects you like numbers agreed before delivery. Put four limits in the PO:

1. Steady-state deviation at nominal speed, as a percentage of setpoint. Around ±3% is reasonable for film and foil with closed loop, ±5–8% for paper, up to ±10% for nonwoven and textile.
2. Deviation during ramp, at a stated acceleration rate. This is usually the worst figure and the one nobody specifies.
3. Deviation across roll change, expressed as scrap metres rather than a percentage: how many metres run out of tolerance after a splice.
4. Recovery time after a defined disturbance — nip closing, splice passing, parent roll change — back inside band within a stated number of seconds.

Then specify the method, since a number without one is an argument waiting to happen: production material, a calibrated reference idler or portable load cell, agreed sampling rate, demonstration at three diameters — full, mid, near core — and a documented roller-tare procedure.

FAQ

Q: How many sensors does a line really need? One per zone where tension error becomes scrap, which is usually two on a slitter-rewinder and three on a coater-laminator. Sensor count follows zone count, not roller count.

Q: One controller per zone, or one multi-channel controller? One multi-channel unit where the zones share a recipe and a single machine PLC; separate controllers where zones are commissioned at different times or may be sold off as separate machines.

Q: Can I buy open loop now and upgrade later? Yes, if the hardware allows it. Specify the torque device and cabling with the future sensor in mind; the expensive part later is mechanical work and downtime, not the cell.

Q: Will better tension control stop the web wandering sideways? No. Lateral position is a separate axis. If your web drifts, you need a web guiding and EPC control system alongside the tension loop.

Conclusion & Next Step

Choosing a system means choosing zones, topologies and acceptance numbers before choosing parts. Map the web path, let the zone map set the topology, spend feedback only where error costs money, size both ends of the roll, and make the supplier demonstrate all four acceptance limits at three roll diameters.

XW Machinery supplies the complete chain — sensors, controllers and magnetic powder clutches and brakes — and sizes them as a matched set rather than as separate catalogue lines. Send us your web path sketch, material and roll diameters for a zone map, a matched specification and a factory-direct quote.