Factory Introduction: What We Make and How
A factory tour video shows aisles, machines and tidy racks. Anyone can walk a plant and leave impressed. The useful question is narrower: at which station would a bad part be stopped, and who signs off when it is. That answer lives in seven process stations and four quality gates, and both can be audited in roughly ninety minutes. Below is the route worth walking, two or three checks per stop, and the numbers that turn an impression into a verdict. Skip the lobby display; ask for the drier log and the first-article board.
01Seven Stations, From Drying Hopper to Packing Tape
The chain runs: resin drying and colouring, injection moulding, trimming and secondary machining, hardware assembly, foam forming and cutting, final inspection, packing. The first four fix dimensions; the last three fix how the case feels in the hand. A visitor need not linger everywhere, but should confirm all seven sit inside one fence. Every process sent outside adds a handoff where tolerances stack and blame gets hard to place.
Chain length shows up as cycle time. Moulding runs 60 s to 120 s a shot; assembly runs 90 s to 150 s a case. That gap forces a buffer of 30 to 60 pieces between them. Look for floor markings and traveller cards at the buffer rather than new machines. A plant with no visible work-in-progress is rarely efficient; it has pushed the inventory into the next building.
02Four Gates Where a Bad Part Should Stop
The gates are incoming material, first article, in-process patrol and outgoing. Incoming needs only two readings: moisture at or below 0.2%, and colour drift within ΔE 2.0 against the master plaque. A first article deserves half an hour of full dimensional measurement, a signature and a board hung beside the press. Without that board a machine is running, not controlled.
Patrol sampling every 2 h, AQL 1.0 on critical dimensions and AQL 2.5 on general appearance, then one more draw before shipment scaled to lot size. A gate exists only if criteria are written down: which dimensions, which gauge, what happens on a miss. Full inspection with no record sheet is not a gate. It leaves no way to date the drift back to a lot.
03A Ninety-Minute Route, Eight to Twelve Minutes a Stop
Enter through the resin store and leave through finished goods; do not double back. In the store, pull the hopper logs and match hopper numbers to the resin. In moulding, check whether every press has its own temperature controller and whether today's first-article board hangs beside it. In secondary machining, ask whether trimming is fixtured and how often fixtures are replaced. That covers about 35 minutes.
Assembly, foam, inspection and packing take the rest. Watch whether station times balance, whether foam has a dust-controlled room of its own, whether gauges carry a valid calibration sticker, and whether carton marks match the shipping list. Ask one question per stop: if this step went wrong, would the next catch it? Where nobody can answer, the chain has a blind spot.
04Three Checks That Take Five Minutes
First, the drying log. Open the last three days and look for gaps in time, temperature and resin grade. A gap means a material change went unrecorded and moisture control becomes a claim rather than a fact. Second, the controller plumbing. Feel the return lines: shared circuits swing mould temperature by 5 ℃ to 8 ℃, and flatness follows by 0.2 mm.
Third, the first-article board. It needs a date, a machine number, measured values and a signature; a missing field means the procedure lives on the wall only. None of this requires instruments, and all of it is awkward to fake. Faking means editing months of records, which costs more than simply doing the work. That is why experienced buyers look at boards before machines.
05Four Numbers That Settle the Question
Press count and clamping range, first-pass yield, rework rate, on-time delivery. Thirty to sixty machines spanning 250 t to 1300 t covers parts through full cases. First-pass yield of 96% to 99% is the normal band here; anything stuck below 96% usually means one station has been firefighting for months. Pull all four from the same period and the same definition, or the set says nothing.
Pin down the rework definition: whole cases or single parts, and whether cosmetic concessions are counted. Ask for six months of delivery data rather than a yearly average. If any of the four has no continuous record, treat it as absent. Cases are bulky and cheap per unit, so one stockout costs far more than the friction of qualifying a new supplier.
| Presses on the floor | 30–60 units |
|---|---|
| Clamping force range | 250–1300 t |
| Cycle time per shot | 60–120 s |
| Moulding hall temperature | 23±3 ℃ |
| Resin moisture ceiling | ≤0.2% |
| First-article sign-off | 0.5 h |
| First-pass yield | 96–99% |
| Patrol sampling interval | every 2 h |
FAQ
Can a shop-floor walk tell you the quality level?
Only partly. The floor shows equipment and takt; quality sits in four record sets: moisture logs, signed first-article boards, patrol intervals and outgoing sampling ratios. Miss two of the four and the plant can start machines, but nobody can prove it holds a process.
Must all seven stations sit under one roof?
Foam and assembly should. They set fit and feel, and outsourced tolerances are hard to trace once they stack. Moulding and machining can sit elsewhere, provided transit packaging and work-in-progress tagging are defined. Otherwise dents and mixed parts surface only at final assembly.
Is more clamping force always better?
No. 250 t to 400 t suits fittings and small cases at lower energy, while 800 t and up is for full shells. Oversized presses stretch cycle time; undersized ones bring flash, sink and drift together. Choose by projected area, not by leaving a spare margin.
What first-pass yield is normal?
96% to 99% in this class. Below 96%, separate tool problems from a process window that is too narrow. Above 99%, check the counting rule: whether parts passed after rework are still booked as first-pass. The definition matters more than the percentage itself.