Cap Tightening Under Control: Magnetic Clutch vs Servo Torque vs Pneumatic, and How to Set Torque by Closure and Leak Risk
Three bottling projects crossed our quoting desk in the same quarter with the same complaint: leaking caps in transit, discovered at the retailer, not at the filler. One was a 500 ml PET sauce bottle with a 38 mm screw cap; one was a 30 ml dropper bottle with a fine-thread closure; one was a 1 L jar with an induction-sealed liner. All three described the machine as a “capping machine” and all three had the wrong tightening method for the closure in front of it. This guide sets out how tightening torque is actually controlled, what each method can and cannot hold, and how to write the torque number into a purchase specification.
The three tightening methods, mechanically
Magnetic clutch. A slipping clutch transmits a preset torque to the cap spindle, and the chattering point is adjusted by hand with a scale or a torque wrench. Torque is limited by friction in the clutch, so it drifts with wear, temperature and the number of cycles. It is inexpensive and mechanically robust, and it is the default on high-speed lines where caps are uniform and the closure is forgiving.
Servo torque control. A servo motor drives the spindle and torque is a controlled output, measured from motor current or from an inline torque transducer. The value is a recipe parameter, changes with a SKU selection rather than with a spanner, and can be logged per bottle for traceability. This is the method to specify when caps are small, threads are fine, or the closing torque is validated in a stability study.
Pneumatic. Air pressure sets the force on the spindle, and torque follows from a combination of pressure and spindle geometry. It is cheap and simple on low-speed inline machines, but it is the least repeatable of the three because air pressure varies with the plant’s compressor loading and with seal wear. On a line sharing a compressor with a case packer, the tightening force at 15:00 is frequently not the force at 08:00.

Comparison table
| Criterion | Magnetic clutch | Servo torque control | Pneumatic |
|---|---|---|---|
| How torque is set | Clutch adjustment by hand | Recipe parameter on the HMI | Air pressure regulator |
| Repeatability over a shift | ±15–25% as the clutch wears | ±3–5% with motor-current control | ±20–30% with shared air supply |
| Repeatability across SKUs | Re-adjust for each cap | Stored per SKU, recalled automatically | Re-adjust regulator per cap |
| Speed range | Up to 200–400 caps/min per spindle | Up to 300–500 caps/min per spindle | Up to 80–150 caps/min |
| Small caps (under 20 mm) | Marginal: clutch resolution too coarse | Good: low torque values resolvable | Poor: overshoot common |
| Fine-thread closures | Acceptable if torque is set high | Best: cross-threading detected by torque rise | Risk of cross-threading on start |
| Traceability | None | Per-bottle torque log available | None |
| Changeover time | 10–30 min of manual adjustment | Under 2 min, SKU recall | 5–15 min |
| Maintenance | Clutch wear parts, periodic recalibration | No wear parts in the torque path | Seal and regulator replacement |
| Price premium vs clutch | Baseline | +35–60% on the capping station | −10 to −20% |
| Best for | Uniform caps, high speed, cost-sensitive | Small or fine-thread caps, regulated products, audit-driven brands | Low-speed inline lines, large caps, non-critical closures |
Torque numbers that actually apply to closures
Cap suppliers publish an application torque range; the machine must be set inside it, and the range is narrower than most buyers assume. The values below are typical starting points for a 25–28 mm neck finish, to be confirmed against the cap supplier’s data sheet and a closure application test.
| Closure type | Neck finish | Typical application torque | What goes wrong outside the range |
|---|---|---|---|
| Screw cap, beverage PET | 26/22 mm | 1.0–1.5 N·m | Below: leaks under pressure. Above: cap skirt splits or neck cracks |
| Screw cap, sauce / dressing | 38 mm | 1.5–2.2 N·m | Below: leakage in transit. Above: liner extrusion and hard opening |
| Fine-thread dropper closure | 18–20 mm | 0.4–0.7 N·m | Below: evaporation loss. Above: cross-threading and pipette damage |
| Flip-top dispensing cap | 24–28 mm | 1.2–1.8 N·m | Below: hinge gap leaks. Above: hinge deformation |
| ROPP aluminium | Neck-specific | Set by roller pressure, not spindle torque | Wrong profile: cracked shoulder or incomplete thread reform |
| Press-on / snap cap | Neck-specific | Set by insertion depth, not torque | Insufficient depth: pop-off. Excess: skirt buckling |
Decision framework: five questions, in order
1 — Cap diameter. Under 20 mm, the magnetic clutch cannot resolve the torque window and pneumatic will overshoot; servo is effectively mandatory. Between 20 and 38 mm, all three are usable and the decision moves to speed and repeatability. Above 38 mm, clutch and pneumatic both work, and the clutch is usually cheaper to own.
2 — Thread quality. Fine-thread or short-thread closures cross-thread when torque is applied before the cap is fully seated. Servo control detects the torque rise of a cross-thread within the first rotation and rejects the bottle; clutch and pneumatic simply tear the thread and pass the bottle downstream.
3 — Output and spindle count. At 12,000 bottles per hour with eight spindles, each spindle sees 25 caps per minute and any method holds torque adequately. At 24,000 bottles per hour with six spindles, each spindle sees 67 per minute and clutch wear accelerates; the recalibration interval drops from monthly to weekly.
4 — SKU count and changeover frequency. More than three cap formats per week on the same line justifies servo purely on changeover time: 20 minutes of manual clutch adjustment per changeover becomes a two-minute recipe recall. Below that frequency, the payback is slow.
5 — Audit and traceability. Regulated products, private-label retailer specifications and any brand that has had a leak claim will usually ask for per-bottle torque records. Only servo produces them natively; retrofitting a torque transducer to a clutch station costs 40–60% of a servo station and still lacks closed-loop control.

Specification checklist before requesting a quotation
Nine items decide the capping station and its price: neck finish and cap diameter, cap material and liner type, cap supplier’s application torque range and the torque-to-open target, target output in bottles per hour, number of spindles the buyer expects, number of cap formats and changeovers per week, bottle stability on the conveyor (diameter-to-height ratio and centre of gravity), whether torque records are required for the customer’s own audit, and the leak test method the buyer will use to accept the line. Without the torque range and the leak test method, two quotations for “a capping machine” are not comparable.
Five selection mistakes that show up as leaks
Buying on caps per minute alone. A machine rated at 200 caps per minute with a clutch will hold torque at that speed only while the clutch is fresh; the same station with servo holds it for the life of the machine. The rating is honest, the comparison is not.
Setting torque by feel at commissioning. Torque set with a wrench on a sample of five bottles, then never re-measured, drifts by 20% within a month. Specify a monthly torque verification with a calibrated torque meter, and write the interval into the maintenance plan.
Assuming a tighter cap seals better. Over-torque is the second most common cause of transit leakage after under-torque: it distorts the liner, cracks the skirt and, on PET, stresses the neck. The correct target is the middle of the cap supplier’s range, not the top.
Ignoring bottle stability. A tall, narrow bottle on a chain conveyor rotates under the spindle, and no tightening method compensates for a bottle that moves. Neck guides and a stable conveyor are part of the capping specification, not an accessory.
Mixing tightening methods across one line. A clutch station feeding a servo station on the same line produces two different torque populations, and the leak test cannot tell which one failed. One method per closure family is the rule.
What each buyer should actually order
Order a magnetic clutch station when caps are 24–38 mm, uniform, and the same format runs for weeks at a time; when output is above 8,000 bottles per hour and the closure is a standard beverage or household screw cap; and when the buyer’s acceptance test is a visual and a random leak check rather than a torque record.
Order servo torque control when any cap is under 20 mm or fine-threaded, when more than three cap formats run on one line, when the brand’s own specification or a retailer’s private-label standard requires torque records, or when the product is a regulated liquid where a leak claim is a recall. A 30 ml dropper bottle at 0.5 N·m is outside the practical range of both clutch and pneumatic control, and this is the single most common case in our files.
Order pneumatic capping only on low-speed inline lines under 3,000 bottles per hour with large caps over 38 mm and non-critical closures, typically household chemicals or industrial liquids where a slight over-torque is harmless. Do not specify it on a line that shares compressed air with a case packer or a palletiser.
Tightening method is not a feature list item; it is the joint between the closure the buyer chose and the leak rate the buyer will be judged on. The capping range and spindle options are listed under capping machines, and how the station is balanced against the filler and the labeller is described under packaging lines. Torque verification procedures and spare clutch parts are covered by the after-sales programme, and buyers who want a capping station reviewed against their own closure data sheets can send them through the contact form.

