Bifilar interwire failure mechanism: partial discharge in the air wedge between enamelled wires; Paschen minimum in air ≈ 305 V, so an interwire stress of a few hundred volts sits at PD onset. Scotchn's ~585 V run is a likely cause. Ceiling for enamel-onl

Trigger: Scotchn (Discord #neuralstan20-testing 2026-09-30) believes the A–B short (#3457) happened at "that 585v mark", a run not yet in my notes. Still to establish: where it was measured, Vpk or Vpp, and whether it came before or after the diode-in captures #3445–#3455. vic_calc gas_discharge, air, 760 Torr (constants A = 15, B = 365, γ = 0.01 assumed):

• 20 µm gap, 400 V: E/N ≈ 797 Td, α ≈ 2,848 /cm, αd ≈ 5.7, M = γ(e^αd − 1) ≈ 2.97 (> 1, self-sustained), Paschen V_b ≈ 347 V; no streamer (αd < 18.5).
• 10 µm gap, 400 V: E/N ≈ 1,595 Td, α ≈ 5,698 /cm, αd ≈ 5.7, M ≈ 2.97, V_b ≈ 307 V.
• Paschen minimum ≈ 305 V at pd ≈ 0.84 Torr·cm (≈ 11 µm at 1 atm). Estimate: the enamel (tens of µm total, ε_r ~3.5) sits in series with the air wedge, so the wedge sees only part of the interwire voltage. This puts PD inception for touching enamelled wires at roughly 500–700 V peak; the exact figure depends on enamel build, which is unknown here. A few hundred volts to ~600 V between adjacent turns is therefore at or near PD onset. At a 20 kHz repetition, PD erodes the enamel cumulatively (millions of events per minute), so failure need not be instantaneous at the peak run. Design rule: in a series-connected bifilar the interwire stress ≈ the half-winding voltage (plus V_D when the diode blocks). For DA-2 (2–5 kV secondary), enamel-only bifilar chokes will run well above PD onset. Use thick insulation (triple-insulated/PTFE), interleaved film, and vacuum impregnation or potting to remove the air wedge, and hipot/PD-test between the windings.