Set the dial to a coin's predicted reading, release the pendulum, and see where the pointer comes to rest. A genuine coin unwinds the dial to the gold PASS band at 0. Readings come from the v2 model, built from Cradle #0001's measured hardware with nothing continuous fitted (21% rms against the readings).
Spacers are 3 mm gauge blocks: the magnet face sits 3S mm above the deck, and the coin gap is 3S − t. The model is shown at 1, 2 and 3 spacers only, matching the readings.
A thicker coin brakes the pendulum more for two reasons. There is more metal (braking goes as conductivity × thickness), and the coin sits closer to the magnet by its own thickness: the N52 ring magnet is 23.4 × 20 mm, its face sits 3S mm above the deck at spacer setting S, and the coin's top face is at gap = 3S − t. Both matter: modelling the coin as a flat slab that ignores this second effect raises the error from 21% to 29%. The coin here is a 32.8 mm 22 ct disc.
to scale · gap = 3S − t, measured by construction