Chris Freedom as-built VIC (2026-11-07 session) — corrections: 3 ppm DI water, 10 mm gap, UX-C2B diodes, chokes C1/C2 on the far leg of the same UY20 core

Source: Chris Freedom message + drawing, Discord #neuralstan20-testing 2026-11-07 (supersedes the 30 ppm / 9 mm / MUR1560 figures I used earlier the same day; extends #3524). AS BUILT for the 2026-11-07 scope shots: switched supply 12 V; Pri 38T; Sec A ~380T 91.5 mH wound over the primary; Sec B ~380T 92.1–92.2 mH wound over Pri + Sec A (all on one bobbin, all clockwise); chokes C1 and C2 each ~380T, 92.1 mH, on their own bobbins on the FAR leg of the same UY20 core (all inductances measured on the core, other windings open). Diodes: UX-C2B (web listings: 8 kV VRRM, 0.35–0.5 A, ~35 ns, Vf ~13.5–14 V; manufacturer PDF not found). Water: 3 ppm DI. Cell: two 304 SS plates 2 x 10 x 124 mm, 10 mm apart on nylon nuts, outer faces insulated with 600 V heatshrink (active area = inner faces, 12.4 cm²). DERIVED (my calc, parallel-plate, no fringing, σ ≈ 5 µS/cm assumed for 3 ppm): C ≈ 88 pF, bulk R ≈ 16 kΩ, self-discharge τ = ε/σ ≈ 1.4 µs (geometry-independent); 18 MΩ·cm water would give ≈ 130 µs. Pri inductance estimate by N² scaling ≈ 0.92 mH (not measured). CONSEQUENCE: C1/C2 are not independent chokes; they link the transformer flux and act as extra ~380T windings whose polarity (aiding/opposing the secondary) and coupling k are unmeasured. Single-core-independent-choke sims from that day (10 mH chokes, 0.72 mH Lp) do not represent this build. Simulator note: the 'core' multi-winding part returned a shorted W1 (0 V across it) in every test on 2026-11-07, so the shared-core model could not be run.