What Repeatable Actually Means
A result is repeatable when the same file, on equivalent stock, under comparable recorded conditions, lands inside the same acceptance band — not when it produces a pixel-identical mark. Defining that band in advance (contrast range, edge quality, absence of char) turns the second run into a pass/fail check instead of a debate.
Questions When Results Differ
When a repeat grid disagrees with its reference, work the variables in order of likelihood:
- Is it truly the same material — same batch, thickness, moisture, and surface preparation?
- Did the optics change — lens swapped, mirrors cleaned or contaminated, focus re-datumed?
- Did the environment move — humidity, ambient temperature, extraction airflow?
- Did the file or firmware change — regenerated grid, altered axis mapping, updated software?
- Is the reference itself trustworthy — was the original sheet fully documented?
Tracking Drift Over Time
Tube output decays and optics foul gradually, so single comparisons can look "close enough" while the trend line is already moving. Logging the same reference cell at regular intervals builds a drift curve: slow decline points to source aging, sudden steps point to maintenance events.
Deciding When to Recalibrate
Recalibration is justified when the confirmation grid leaves the acceptance band, when any controlled variable on the sheet can no longer be reproduced, or when the material supply itself changes batch. Recalibrating on every whim destroys the baseline; never recalibrating guarantees silent failure. The ledger — not intuition — should make the call.