Micro screws have a distinctive profile: individual weight measured in milligrams, batches running to hundreds of thousands of pieces, unit value too low to justify piece-by-piece inspection, and yet a batch problem stops the customer's entire assembly line. Quality therefore cannot come from sorting after the fact; it has to come from stability inside the batch. We produce stainless micro screws from M0.6 to M3 and break the closed loop into executable steps.
Control starts with defining the batch. For micro screws a batch should not be a production date but the combination of wire heat number, heading machine, thread rolling die, heat treatment run and plating bath. Wire from one heat spread over two days is still one material source; a new wire heat, even at the same size, starts a new batch. The batch number must resolve to all five items, otherwise an anomaly at the customer cannot be scoped and the whole store is quarantined - a loss far larger than the record keeping.
Incoming stainless wire needs more than a mill certificate. Verify grade and condition (annealed, hard drawn or cold drawn), diameter and tolerance, surface condition for draw marks and rust spots, and hardness within the heading window. Too hard and heads crack during heading; too soft and threads do not fill out, failing the gauge. A short trial heading run shows head formation and die wear more reliably than any report. Store wire off the floor and away from moisture, since damp seasons leave water-stain rust on the surface.
Heads form by cold heading and threads by rolling, and both set dimensional stability. Heading depends on stroke, feed length and die clearance: a longer feed makes a larger head diameter, while excessive clearance produces flash or oval heads. Rolling depends on die parallelism, feed rate and centre height: wrong centre height yields tapered threads or flattened crests. Keep reference samples from each setting, and run a full first-article dimensional check after every die change instead of waiting for the patrol inspection.
Most stainless micro screws are not through hardened, but stress relief annealing is common to remove cold work and restore toughness. Insufficient temperature or soak leaves screws brittle in assembly; excessive temperature grows the grains and lowers thread strength. The greater risk is hydrogen from plating: acid pre-treatment introduces hydrogen, and incomplete baking lets a micro screw fracture suddenly after hours under load, with almost no deformation at the break. Baking time and temperature must be recorded, not estimated.
Coating consistency is where micro screws most often fail. In one bath, parts at the rack edge and at the centre can differ in thickness by twenty to thirty percent; current density at thread crests builds extra thickness until the go gauge no longer passes. Control rack density, analyse and agitate the bath, and sample thickness and adhesion from every lot. Where conductivity or solderability matters, excessive thickness raises contact resistance, so state the requirement on the drawing rather than assuming a standard coating.
Sampling separates critical from general characteristics. Critical items are thread go/no-go, head diameter, head height, shank diameter and overall length; general items are head appearance, recess depth and concentricity. Sample size follows lot size and historical defect rate - generous for large, stable lots, tightened after a single nonconformance. Match the instrument to the feature: micrometer for shank, profile projector or vision system for head and recess, gauges for thread, and optical sorting as a full-check aid at very small sizes.
The outgoing report should carry measured data, not the word conforming. Items include appearance, dimensions, thread, material certificate, coating thickness, salt spray where specified, and cleanliness. Agree acceptance rules in advance: a critical defect - fracture, blocked thread, mixed parts - rejects the lot; major defects trigger tightened sampling and re-judgement; minor defects are handled against an acceptable quality level. Mixed parts are the common failure at micro size, so confirm size and material before packing, using magnetic or spectral sorting if needed.
The loop closes with records and retention samples. A modest sample per lot suffices if it covers every size and coating condition in that lot, kept for the period agreed with the customer, typically at least a year. Records link wire heat, machine, die number, process parameters and inspection data to the shipped lot. Producing this set during a supplier audit is worth more than any assurance. We keep one file per order and ship the inspection report with a batch cross-reference, so an anomaly can be scoped within two hours.