VIC design session v66: reasoning

This sweep did not do what I set out to do, and that's the headline. The intent was to force L up to 1.5–2mH with C_eff cut to 4 cells so f_r would drop into the audio band. Instead every top candidate reports 2×0.1mH chokes, 8 cells at 1.37nF, and f_r pinned at 303.9kHz — 3× above my 100kHz ceiling. So the "winners" all carry the same VIOLATION. The optimizer chased the CGDE film-onset reward (110–126 Td, squarely in the 3–6 eV cold-chemistry window) and the high cell voltage that comes with it, and paid for that with an out-of-band resonance. The scoring is rewarding physics I want but on a circuit I can't practically drive.

The trade-off is legible across #1→#5: as I pile on turns and thinner gauge to chase inductance/current, coil Q collapses (886→150) while tank Q barely moves (26.4→23.0). That tells me tank Q here is capped by the 14.1Ω cell series R plus bulk water R, not by the coil — so throwing copper at it is wasted. Cell voltage tracks tank Q directly (316V→276V), and film Td tracks cell voltage. Current restriction and Q are the same lever; frequency is the casualty.

I'm provisionally adopting #1 for its numbers, but flagging it as non-compliant — it is a target, not a build.

Next session: actually enforce L=1.5–2mH and C_eff=4 cells as hard constraints before scoring, verify f_r lands 40–70kHz, then re-sweep turns/gauge to recover cell voltage and keep the film inside the Td window at the lower frequency.