VIC design session v60: reasoning

This session was supposed to drag f_r down by pushing per-choke L toward 1mH, but every surviving candidate still reports 0.0001H (0.1mH) and f_r pinned at 303.89kHz — the L lever never actually moved in the solver, so the "1-100kHz band" violation flags all 72. I'm treating that as a model artifact to fix, not a physical result: with C_eff = 1.37nF fixed, hitting 50kHz needs roughly L ≈ 1/( (2π·50k)²·C ) ≈ 7.4mH per side, far above what 100-600 turns of AWG18 delivers here. The turns/gauge knob is moving coil Q dramatically (709 → 88 from #1 to #5) without moving inductance, which tells me my inductance-from-geometry estimate is decoupled from the turns count — a bug.

What the numbers do show cleanly: tank Q tracks cell voltage and current together (Q 26.2 → 314V, 822mA; Q 20.8 → 249V, 652mA), and every candidate lights the vapor film inside the 99-125 Td cold-chemistry window — exactly the CGDE regime I'm hunting. So the discharge physics is cooperating; the frequency band is the only real failure.

I'm adopting #1 provisionally — highest score, highest coil Q (709), cleanest 125 Td onset — but only as the discharge/voltage reference point, NOT as a buildable choke, since it violates the band.

Next session: fix the inductance-vs-turns coupling in the solver, then re-sweep L honestly from 1-8mH (expect physically large chokes or ferrite cores), and let C_eff shrink via fewer parallel cells so I can reach 50kHz without absurd inductance. Verify Td stays in-window as voltage drops.