VIC design session v37: reasoning

This sweep did not do what I set out to do. I wanted the in-band 5–7.4mH branch that resonates the 1.37nF cell at 40–70kHz, but the scorer instead crowned the low-L branch — every top candidate sits at 2×0.1mH chokes and rings at 303.9kHz, four to seven times above my practical 1–100kHz ceiling. The optimizer is being seduced by the CGDE film diagnostic: higher tank Q lifts cell voltage (303V at Q=25.2 down to 243V at Q=20.2), which lights the vapor film harder (121 Td down to 97 Td) and all five land inside the 3–6 eV cold-chemistry window. So the score is tracking film chemistry, not buildability, and it happily trades my frequency band away to get another few volts across the 0.1mm film.

The real trade-off is clean here: coil Q buys cell voltage and Td, but it comes almost entirely from turns/gauge (candidate #1's 200-turn AWG18 gives coil Q 354.6 vs #2's 400-turn at 177.3), while tank Q is pinned by the 14.1Ω cell series R. Current restriction stays modest — 635–793mA — because these tiny chokes barely limit at resonance.

I'm not adopting any of these for the bench. #1 is the "best" number but it violates the band, and I distrust a 303kHz design driving real iron. If I must pick, #1 documents the ceiling. Next session: force L back into the 5–7.4mH in-band region and re-let the scorer choose there, and add a hard band-violation penalty so out-of-band candidates can't win. Then sweep turns/gauge within that band to recover ~300V film at buildable frequency.