The customer's gear inspection package, assembled from your records for your quality manager to sign.

An inspection requirement can come from the print, the purchase order or a specification either one names. The evidence sits all over the building: profile, lead, pitch and runout charts from the analyzer, a microhardness traverse for the case, surface and core hardness, temper etch or particle inspection sheets, the mill's certificate, and the traveler that ties them to a serial or lot.

For the test, AI reads copies of those records and assembles the package in your layout: a cover naming the print, revision and stated quality class, each record listed with its source, and a list of what the requirement asks for that the file doesn't hold. It copies measured values and passes no verdict on them. Whether a gear met its class is for your quality manager to say.

A chart belongs to the method that produced it. Under AGMA 2000-A88 a trace passed if it stayed inside a K-shaped band. ANSI/AGMA 2015-1-A01 uses line-fit analysis and reports total, slope and form deviation, as Edward Lawson explained in Gear Technology in 2004. What a job needs depends on its grade: Louisiana's movable bridge chapter lists cumulative and single pitch for the low accuracy group, adds total profile and lead for the medium group, and adds slope and form for the high group. So the file notes which document each chart was run to, and lists any chart your quality manager's checklist calls for that isn't there.

The instrument has a paper trail too. ISO 18653 addresses traceability, calibration intervals and sources of measurement error. Robert Frazer and Steven Wilson observed in Gear Technology in 2010 that uncertainty is seldom considered until a dispute, usually a customer's machine against a supplier's. Lawson wrote that a master gear without a current report from an accredited laboratory, stating its uncertainty, is of no value for calibrating a measurement process. Where a customer wants that evidence, the package shows the master's report and date, or marks it missing.

Case depth depends on whose definition you use. In Gear Technology for January/February 2020, Robert Errichello and Andrew Milburn put the European effective case depth at 550 HV and the US one at 50 HRC, which reads deeper, with no easy conversion between them because the hardness gradient varies by gear, steel and process. Some laboratories report both. They place separate readings at mid-height of the flank, at the root fillet and at the tip, take core hardness on the root circle, and note that a typical drawing states depth after final machining. So each case depth on the file shows its definition, location and stage, and the list picks up any of those a heat treat sheet omits.

Some records describe and leave the verdict open. ANSI/AGMA ISO 14104-A17, on surface temper etch after grinding, replaced ANSI/AGMA 2007-C00. MPMA's catalog says it classifies what an etch shows and holds no acceptance or rejection criteria, and the ISO scope leaves out nitrided parts and stainless steels, so the limit has to come from the print or the customer's spec. AGMA 923-C22 works the other way, giving acceptance criteria by heat treatment and grade number, so a metallurgical record only makes sense beside its grade. A 2010 Heavy Movable Structures paper shows one buyer pattern: rotating parts ultrasonically examined before machining and magnetic particle examined after, gearing checked for lead, profile, tooth spacing and runout, with some authorities wanting Q8, some Q9 and some Q10.

What goes to the customer is a decision. A 2008 paper at the same symposium argued that submittals for bridge reducers should stop at outline drawings, gear rating calculations and material certifications for gearing, since proprietary information is at stake. So the draft separates what the requirement names from what merely sits in your files. Your reviewer, usually the quality manager, checks that the right job's records were found, every missing item is real and nothing was called good or bad. Read this as a summary of public papers and standards listings, not as engineering advice.

After a yes, each job nearing shipment gets a drafted file for about 30 days, with time left to go find a missing chart.

Your quality manager reviews every file, decides what each measurement means and signs it. Nothing leaves without that signature, and any further job is its own decision at its own price.