How to Calibrate Weights Before They Leave the Factory
Every weighted band leaves the mould with a number printed on it. Until that piece has been placed on a verified scale, the number is a claim rather than a measurement, and the gap between the two is what buyers discover in a studio three weeks later. Calibration before dispatch is not a laboratory ritual performed once a year on a certificate; it is a short, repeatable bench routine that turns a nominal label into an honest one, and it takes less time per unit than most packing steps.
A Nominal Weight Is a Claim Until It Is Weighed
Three quantities need to be kept apart: the nominal mass printed on the band, the actual mass of the piece as built, and the tolerance the buyer has agreed to accept around the nominal figure. Most disputes arise because only the first was ever written down. A 2 LB pair that arrives at 1.94 LB is not a broken product, it is an unmanaged tolerance, and the difference matters because a studio counts repetitions against load and a retail buyer counts against a specification sheet.
Defects should be sorted before the first carton is packed, not after a rejection. In practice three classes cover almost everything: critical defects, which include a label mismatch larger than the agreed tolerance band, a missing or displaced stainless iron core, and any strap that fails a 200 N pull test; major defects, which include seam splits, flash over 1 mm, uneven wall thickness visible at the edge and a barcode that will not scan; minor defects, which include surface marks, mould release lines that do not affect grip, and print misregistration under 0.5 mm. A workable sampling plan sets the acceptance number at zero for critical, and applies an acceptance quality limit of 1.0 for major and 2.5 for minor under ISO 2859-1, inspection level II.
Reference Masses and Scale Division on the Bench
The bench needs one verified reference set and one scale whose resolution matches the tolerance being judged. Reference masses are classified by OIML R 111-1:2004, which defines classes E1, E2, F1, F2, M1, M1-2, M2, M2-3 and M3, from the most accurate downwards. A production floor does not need class E1. Class F1 or F2 covers the master check mass used to verify the scale at the start of a shift, and class M1 is acceptable for the daily working check provided its calibration certificate is still in date and the mass has no visible damage. Each certificate states the maximum permissible error for that individual mass, and that figure, not the class name, is what enters the record sheet.
Resolution matters more than brand. A useful rule is that the instrument should resolve at least one tenth of the tolerance it is judging, read as a division value rather than as a display digit. If a pair of 1 kg ankle weights carries a tolerance of plus or minus 20 g, a scale with a 0.5 g division is adequate; if a 0.5 kg band carries plus or minus 5 g, the bench needs 0.1 g. Note the verification scale interval e on the calibration sticker and keep the load between one tenth and the full capacity, because a 30 kg platform scale asked to weigh 400 g will round away most of the error being investigated. Check the scale with the master mass before the first piece and after the last piece of the shift; if the two readings differ by more than one division, stop, re-zero and investigate before any result is signed.
Sampling: First Article, Then a Defined Share of the Lot
Weighing every piece forever is expensive, and weighing five pieces from a thousand-unit lot is theatre. A two-stage plan resolves the tension. In the first article stage, the first 200 pieces of a new mould, a new core batch or a colour change are weighed one by one, and the run only continues when every piece falls inside the band. In the steady state stage, draw 32 pieces from a lot of 1,000, 16 from a lot of 500 and never fewer than 5 from anything smaller, spread across the top, middle and bottom layers of the packed carton rather than taken from the first row the operator can reach.
Two extra samples close the loop. After packing, re-weigh 8 finished units because the bag, tag, band and fitment add mass that the declared gross weight must reflect, and a carton declared at 20 kg gross for four 10 LB triangle rings is checked against that figure at the forwarder's desk. After any tooling service, repeat the first article stage in full, since a repaired mould cavity rarely returns to exactly its previous volume. Our production and testing lines run the first article stage in line rather than in a separate room, so a drifting cavity is caught within the first hour rather than at the end of the order.
Tolerance Bands in Grams and Parts per Thousand
Cast silicone around a steel core has two sources of variation that never disappear: cavity volume changes as the mould wears, and the core sits a fraction of a millimetre away from its nominal position in each shot. A realistic band therefore has to be expressed relative to nominal mass and then converted to grams on the sheet. For pieces above 500 g, twenty parts per thousand is achievable and fair, which means plus or minus 20 g on a 1 kg piece and 40 g on a 2 kg pair weighed together. For pieces under 200 g, where the core is a larger share of the total and the silicone skin is thin, thirty parts per thousand is more honest, so a 0.5 kg band carries plus or minus 15 g. Tighter than ten parts per thousand on a cast part invites either scrap or a hidden rework step, and neither is visible in a quotation.
Dimensional checks belong in the same routine, because a core that has shifted sideways changes the balance of the piece without changing its mass. Measure the strap length, the band width at 55 mm, the thickness at the seam and the core position against the moulded centreline, and accept a 3 mm deviation at most on a 4 LB ring. Temperature and humidity shift the reading even when the product is perfect: weigh in a room held at 20 degrees Celsius plus or minus 2 degrees and 50 per cent relative humidity plus or minus 10 per cent, let pieces from a cold container stand for 24 hours and pieces straight from the press stand for 30 minutes, and keep the bench away from door draughts and air conditioning outlets, since moving air on a 0.1 g division scale is enough to make a stable reading impossible.
The Record Sheet That Travels With the Carton
One sheet per lot, one line per check. It carries the lot code, the SKU and nominal mass, the scale identifier and its verification scale interval, the reference mass identifier with the certificate number and expiry date, the room temperature and humidity at the start and the end of the session, the number of pieces weighed in the first article stage with the highest and lowest readings, the sample size and mean deviation in the steady state stage, the defect counts by class, the operator name, the date, and a disposition line signed by whoever releases the carton. Photograph the display for the first article piece and for any reading at the edge of the band, because a photograph settles an argument faster than a transcription.
Retain the sheet for three years and send a copy with the shipping documents. When a buyer later questions a single pair, the sheet shows what was measured, on which instrument, against which reference mass, at what temperature, and who signed it. That is the difference between a complaint that is answered in an email and one that ends in a credit note. Weight ranges and piece counts for each SKU are listed on our product pages, and questions about a specific tolerance band can go through our contact page before the purchase order is written.
Frequently asked questions
What weight tolerance should I specify for a silicone weight?
For cast pieces above 500 g, twenty parts per thousand of nominal mass is realistic, which is 20 g on a 1 kg piece. Below 200 g, where the steel core is a larger share of the total, thirty parts per thousand is fair. Anything tighter than ten parts per thousand on a cast part usually means scrap or a hidden rework step.
Do you calibrate every piece or sample the batch?
Both, in sequence. The first 200 pieces of a new mould or core batch are weighed one by one, and afterwards 32 pieces are drawn from a 1,000 unit lot, spread across the top, middle and bottom of the packed carton. Eight finished units are re-weighed after packing so the declared gross weight is honest.
Which reference weights and scale division does the bench need?
A class F1 or F2 master mass verifies the scale at the start and end of a shift, and class M1 is acceptable for the daily check while its certificate is valid. The scale should resolve at least one tenth of the tolerance being judged, so 0.1 g for a 0.5 kg band and 0.5 g for a 2 kg pair.
How is AQL applied to silicone fitness products?
Critical defects such as a label mismatch beyond tolerance or a displaced core carry an acceptance number of zero. Major defects such as seam splits, flash over 1 mm or an unscannable barcode use an acceptance quality limit of 1.0, and minor cosmetic marks use 2.5, both under ISO 2859-1 at inspection level II.