The main event: watching a bare relay coil's collapsing field with no freewheel diode to tame it.

Setup — 24 V coil, diode removed, two 10x probes straddling the disconnect (CH1 supply side, CH2 coil side). The agent armed a single-shot on CH2's falling edge and waited. The first pulls just clipped: the kick blew past ±130 V, then −304 V, railing the capture window each time. Two things cracked it — breaking from the minus side of the bulk caps took the supply capacitance out of the loop so the coil rang free, and then stepping the vertical coarser and re-arming until the whole transient fit at 100 V/div.

There it was: a −436 V spike with a +240 V overshoot — a ~676 Vpp ring off a 24 V rail, roughly 18× the supply — then an exponential bleed-back over ~1 ms. Drop a freewheel diode back in and the same coil clamps to a tidy ~24.7 V. That's the whole argument for the diode, in one screen.

Waveform CSVs for both channels and a machine-readable capture summary are attached under Files

.