What drifts, and how fast.
Each ageing process has its own clock. Tubing is the fastest: silicone hardens over winters, cracks at the instrument spigots, and slips off during the re-panel — most leaks are born in the workshop, not in the air. Baro sensors drift slowly, on the order of a hectopascal a year — eight metres of altitude — which is invisible day to day and decisive over a decade of badge claims. Batteries fade by calendar as much as by cycles. And the magnetic environment does not drift at all; it jumps, every time something ferrous or current-carrying joins the panel.
The leak test is the whole ballgame.
Most instrument complaints that reach our bench are plumbing. The test that finds them is beautifully dumb: pull a gentle suction on the static line until the altimeter shows a test height, close the system, and watch. A tight system holds; a leak decays back toward ambient along an exponential, fast at first, and the loss in the first minute is your number. One warning worth a paragraph of its own: never blow into a port by mouth. A pressure sensor that measures centimetres of altitude resolves fractions of a millibar and a lung delivers a hundred; a well-meaning puff is a common way instruments arrive at a service bench dead.
The leak-test bench.
Slide the leak up and watch the decay steepen — the marker sits at the one-minute mark, where the pass/fail conversation happens. Notice the trap in the exponential: a marginal leak loses its metres slowly enough to look almost flat for the first seconds, which is why the test is timed with a watch and not judged by eye. The 30 and 100-metre bands are teaching thresholds; your maintenance programme's numbers govern.
The baro sensor and the chamber.
Drift cannot be checked against itself, so once a year the instrument meets a reference. The everyday version costs nothing: altimeter against field elevation at today's QNH, error written down, trend watched year on year — one season's eight metres is noise, three seasons in the same direction is a sensor talking. The formal version is a pressure chamber sweep against a laboratory reference, which is how IGC-approved recorders keep their calibration certificates valid and how our own bench in Celje spends its mornings. If your recorder's certificate has expired, the badge claim it signs one day is arguing with a piece of paper.
In your cockpit.
The season's first flight is the last station of the annual check. Fly it as a test flight on a calm day: airspeed against GPS ground speed both directions, altimeter against a known circuit height, vario quiet in steady cruise, heading against the runway you are lined up on. Ten deliberate minutes, and every instrument has either confirmed the winter's work or filed a complaint while the fix is still a hangar job. The plumbing this all rests on is the pitot-static article, and the data that ages alongside the hardware is the database hygiene article.
Every unit leaves through a calibration chamber — the reference your annual check compares against.
The airdata box on the service bench gets a fresh calibration certificate, not just a pass.
Sensors chosen for drift you can measure in years, and a service path when the years add up.