A preventive maintenance record is not a calibration log

This blog has already covered what a calibration record owes Clause 7.6 — a traceable standard, a documented basis when no standard reaches far enough, a visible status on the equipment itself. A quieter clause sits two sections earlier in the same standard, governing a question calibration doesn't ask at all: not whether a piece of equipment measures accurately, but whether it keeps working the way it's supposed to. Quality files that fold both duties into one maintenance-and-calibration binder are usually missing whichever half doesn't get a due-date reminder.

Two clauses, two different failure modes

ISO 13485:2016 Clause 7.6 governs monitoring and measuring equipment — gauges, scales, sensors, anything whose job is to produce a trustworthy reading. Clause 6.3, Infrastructure, governs something broader and less glamorous: the buildings, workspace, utilities, and process equipment — hardware and software — that a manufacturer needs to make conforming product at all. A scale that reads five grams high has failed a calibration question. A molding press whose heating element is degrading, or an HVAC system that's stopped holding a cleanroom's specified pressure differential, has failed a maintenance question, and no amount of calibrating the thermometers inside it will catch that kind of failure, because the instruments can be reading correctly the whole time the equipment around them is drifting out of the condition it was qualified in.

The duty is conditional, not blanket

Clause 6.3 doesn't ask a manufacturer to write a maintenance procedure for every object on the floor. It requires documented requirements for maintenance activities, including their interval, specifically when those activities — or their absence — can affect product quality. That's the same shape of trigger this blog has already traced in Clause 7.5.6's own conditional test for process validation: the clause doesn't scale with how expensive a piece of equipment is, it scales with whether skipping its upkeep could put product at risk. A conference-room projector doesn't need a documented PM interval under 6.3. The press, the autoclave, the cleanroom's air-handling system do, because what happens inside them if maintenance lapses is exactly the kind of quality-affecting drift the clause exists to catch before it reaches a device.

An interval is not the same claim as “maintained as needed”

The clause's own wording names the interval as part of what has to be documented, not just the activity. A maintenance SOP that says equipment will be serviced “as needed” or “per manufacturer recommendation, if available” has described a philosophy, not a requirement a floor technician can be checked against. The working version states a stated frequency — hours run, cycles completed, or calendar time, whichever the equipment's own failure pattern actually tracks against — and a record showing whether that interval was met, the same discipline a calibration due-date already imposes on the instrument sitting next to the equipment being maintained.

Maintenance is what keeps a qualified process qualified

The two clauses meet at the equipment a process validation already depended on. A validated process is qualified against a specific piece of equipment operating within specific parameters, and Clause 6.3's maintenance duty is what's supposed to keep that equipment inside the envelope the qualification assumed, in between the calibration checks that confirm its instruments are still reading straight. A press that's never recalibrated because its instruments still check out, but whose mechanical wear has quietly shifted its actual cycle time outside the range the process validation qualified, is a maintenance failure a calibration program was never built to catch. Treating a shared PM-and-calibration spreadsheet as one undifferentiated duty is how that gap goes unnoticed until a lot fails downstream for a reason the maintenance log should have flagged first.

Where this meets the file

A preventive maintenance record needs its own field structure — which equipment triggers the 6.3 duty, at what documented interval, and what happened at each service event — kept distinct from the calibration record's own traceability and status fields, rather than one combined log that answers neither clause completely. A maintenance-record template built around that structure, alongside the launch catalog's calibration and process-validation templates, is previewed there now. If your program runs maintenance and calibration off one shared log and it's worked for you, the shelf takes that correction directly.