A contract packer lives or dies by the number of box sizes it can run in a single shift. Where a brand owner might run two or three carton formats all year, a co-packer swings through dozens of customer programmes, and every new SKU is a stop, a tune, and a lost quarter-hour. The machine that wins this job is not the fastest on paper — it is the one whose changeover disappears into the noise of the shift.

This article explains how a sub-5-minute changeover is actually engineered across a catalogue of 50 or more SKUs. It covers the daily rhythm of a contract packing line, the three machine classes and what each costs to retool, the six minutes that make up a real changeover, and how SMED discipline and PLC recipe memory turn availability back into shipped cases. It closes with flute-grade adaptation and the alarms that most often stall a retool.

The Short Answer

Across 50-plus SKUs the changeover, not the nameplate speed, decides your throughput. A fixed-format erector can lose 45 to 85 minutes every time it switches box size; a servo-memory machine collapses that to under 300 seconds, most of it a three-case trial run.

The envelope that matters is L 200–600 mm, W 150–400 mm, H 100–400 mm. Inside it one machine serves the whole book of business; outside it you are forcing a tool change that should have been a different machine class. At six to ten changeovers a day, the difference between eight-minute and five-minute swaps is 540 to 900 cases per shift of recovered output.

A Contract Packer’s Real Day

Forget the brochure speed. The duty cycle on a co-packing line is a sequence of short runs: batches of 500 to 5,000 cases, each followed by a retool. A typical plant executes six to ten changeovers per shift, and the batch size, not the machine speed, sets how often the line stops.

That rhythm changes the optimisation target. At 40 CPM a run of 500 cases is finished in 12 minutes, so the changeover beside it is no longer a rounding error — it is half the cycle. A line that changes ten times a day at eight minutes each spends 80 minutes, or the better part of a shift, not making cases. Cut that to five minutes and the same day returns 30 minutes of run time, roughly 1,200 cases at 40 CPM.

Three Machine Classes and Their Retool Cost

The market splits into three tiers, and the changeover cost scales with how much of the adjustment is mechanical versus recalled from memory.

Fixed-format erector. Tooling is physically dedicated to one box. Switching sizes means swapping mandrels, rails and glue heads, then re-validating squareness — typically 45 to 85 minutes of labour and a 200 to 400 USD tooling kit per size. cheapest to buy, most expensive to live with once your SKU count climbs past a dozen.

Adjustable erector (tool-free handwheels). Rails, flap guides and cup mounts move on handwheels and lever locks, no spanners. A competent operator completes a retool in 12 to 25 minutes, with no parts to buy, so the cost is pure labour: roughly 25 to 50 USD at 12 USD per hour. This is the workhorse for 10 to 50 SKUs.

Full-servo memory machine. Every axis is driven and the whole geometry is stored as a numbered recipe. Changeover is a HMI pick plus a three-case trial: under 5 minutes, almost none of it manual. The premium is in the capital — typically 30 to 60 per cent above the fixed machine — but at 50-plus SKUs the recovered minutes pay it back within the first year of mixed-volume work.

Breaking Down the Five Minutes

A real changeover is six operations, and the sub-5-minute claim only holds if each is designed to be fast. The breakdown below is what a servo-memory machine achieves; a handwheel machine simply does the same steps more slowly by hand.

Box-size adjustment (L/W/H rails). Servo axes drive the case magazine, compression section and flap folders to the new dimensions in 15 to 30 seconds; on a handwheel machine this is 5 to 10 minutes of cranking and measuring.

Suction cup position. Pick heads must reach the new blank geometry. Quick-change cup plates swap in under 60 seconds on a servo unit; manual re-bolting takes 3 to 6 minutes.

Flap guide rails. The minor and major flap folders reposition automatically in 10 to 20 seconds; manually, 2 to 4 minutes of loosening and setting.

Glue line position. The hot-melt head shifts to the new seam and re-registers the photo-eye in 10 to 20 seconds; manually, 2 to 4 minutes plus a purge.

Recipe call-up. The PLC loads the stored geometry, vacuum profile and glue grammage for the SKU in under 10 seconds; on non-memory machines this step does not exist and every parameter is set by eye.

Three-case trial run. Three blanks are erected and checked for squareness and seal before the run is released — 60 to 90 seconds on every machine class, the one step that should never be skipped.

SMED in Practice

The discipline that gets you under five minutes is SMED — single-minute exchange of die. Its core move is sorting internal operations (machine must be stopped) from external ones (can be done while running), then eliminating the internal set.

Externalise what you can. Pre-kit the next SKU’s cup plate and glue profile on a shadow board beside the line while the current run is still packing. A contract packer doing ten swaps a day should never start a changeover hunting for parts.

Quick-change tooling. Magnetic and cam-lock cup plates, tool-free flap guides and colour-coded mandrel inserts turn a six-bolt job into a 30-second swap. This is where most of the handwheel machine’s time actually goes.

PLC recipe memory. A full-servo machine stores 50 to 200 recipes, each holding axis positions, vacuum setpoint, glue grams per case and compression dwell. Call-up is under 10 seconds and removes operator judgement from the loop — the single largest availability gain for a 50-plus-SKU book.

Specifying the Box Envelope

The changeover promise is only real inside the machine’s mechanical envelope. The range below is the working window our engineers quote; push past it and the retool stops being a recipe and becomes a rebuild.

When a customer’s carton falls outside this window — a very long 700 mm display tray or a shallow 80 mm mailer — the right answer is a second machine class or a custom mandrel, not an over-travelled servo that drifts out of squareness. Sizing correctly before purchase is what keeps the sub-5-minute claim honest.

OEE and the Cost of Changeover

Changeover time attacks availability directly, and availability is the first factor in OEE. Overall Equipment Effectiveness multiplies availability, performance and quality, so every minute spent retooling is multiplied through the whole shift.

OEE = availability × performance × quality
Availability = (planned time − changeover − unplanned downtime) ÷ planned time

Take a shift of 480 planned minutes with six changeovers at eight minutes: that is 48 minutes lost before a single fault. Availability before performance and quality is already 0.90. Drop changeover to five minutes and the same shift recovers 18 minutes, lifting availability toward 0.94 and, at 0.95 performance and 0.99 quality, pushing OEE from 0.85 to 0.88 — a 3-point swing earned with zero new capital. For a line packing more than 200 SKUs, a Single-Piece Carton Erector with recipe memory is the cheapest way to bank those points.

Running E, B and C Flute on One Machine

A co-packer rarely controls the board grade, so the same erector must handle E, B and C flute without a changeover tax. The variables that move are the vacuum demand of the blank and the compression dwell, because thinner boards need a gentler, cleaner pick.

The practical rule is to store one recipe per flute-and-size combination, so switching from a 300 mm B-flute mailer to a 500 mm C-flute shipper is still a single HMI pick. A dusty, high-recycled board raises the vacuum demand by 0.05 to 0.10 bar, which is why the recipe should carry a vacuum margin rather than a tight setpoint.

PLC Alarms That Stall a Retool

Most changeover delays are not mechanical — they are the machine refusing to run until a fault clears. The four alarms below are the usual suspects when a swap overruns.

Routine care keeps those alarms away from the changeover window. Every shift: confirm glue is clean, the nozzle is not weeping, and suction cups are dust-free. Every week: purge the nozzle, verify vacuum at the pump, and check belt and encoder tension. Every month: torque the main drive belts, inspect the compression section for board dust, and verify squareness against a reference case. Every quarter: replace picking-head suction cups and clean or replace filters. Every year: re-tension the drive chain, replace the nozzle assembly, and calibrate servo datums and temperature controllers. Lines in this rhythm typically keep availability above 98 per cent across three shifts.

Choosing for Your Line

The decision is driven by SKU count more than by speed. Below a dozen sizes a fixed or adjustable erector is the rational buy; between 10 and 50 SKUs the handwheel machine earns its keep; above 50 SKUs, or wherever batches fall under 1,000 cases, the full-servo memory machine is the only class that protects availability. The high-volume twin-lane duty is a different problem and is better served by our Double-Head High-Speed Case Former, while tray and three-piece work is covered by the Three-Piece Carton Erector.

Specifying the envelope and the recipe count before purchase is the step that decides whether your changeover is a five-minute formality or a shift-eating event, and the sizing questions are answered in our FAQ.

FAQ

Can a fixed-format machine ever hit five minutes?
Only if every size shares the same footprint family and you pre-stage tooling; realistically it stays in the 45-minute band. For 50-plus SKUs it is the wrong class.

Does recipe memory need an encoder re-zero after a fault?
After a power loss or a E-33 alarm, yes — re-home the axes and re-verify squareness on three cases before释放 the run. The recipe holds positions, not the physical datum.

What batch size makes servo memory pay back?
Roughly any mix averaging under 1,000 cases per SKU per run. Above that, the machine speed matters more than the changeover, and a simpler class is fine.

Why keep a three-case trial if we are in a hurry?
Because a skewed case that jams the conveyor costs 5 to 15 minutes to clear — far more than the 90 seconds the trial consumes. The trial is never the bottleneck.

Ready to cut your cost per case? Send your carton specifications — length, width, height, flute grade, target cases per minute and annual volume — to [email protected] or WhatsApp +86 13681839278. Our engineers will return a changeover-time model and a machine sizing sheet within 24 hours, with no obligation.