VIC design session v35: reasoning

Lab note — VIC resonant sweep, RANDLES-CELL + vapor-film diagnostic.

The premise this session was to abandon the phantom 200nF sizing and push L up toward ~7.4mH to land f_r=50kHz on the real 1.37nF cell. The scorer did NOT do that. Every top candidate still carries the 0.0001H chokes and sits at 303.89kHz — three times outside the 1-100kHz practical band. So the headline result is a negative one: the optimizer preferred high-frequency, high-Q solutions and never actually adopted the heavy inductor I asked it to sweep. I need to check whether 7.4mH was even present in the candidate pool or got pruned on coil Q/DCR before scoring.

What the ranking teaches: score tracks the vapor-film Td almost linearly. #1 wins at 128 Td (321V film) with tank Q=26.8 and 841mA; going down the list, thinner gauge and more turns bleed coil Q and cell voltage, dropping Td and score together. So tank Q buys cell voltage buys CGDE onset — but it rides on a frequency I can't drive cleanly. Current restriction (the whole point of the series chokes) is barely exercised at 0.1mH; these chokes are current limiters in name only here.

I'm provisionally adopting #1 — heavy AWG18, highest coil Q (1773), squarely in the 3-6 eV window — but flagging it as band-violating and therefore not buildable as-is.

Next session: force L=1-10mH into the pool explicitly, re-verify f_r lands 40-70kHz, and re-check that fatter chokes keep enough Q to hold ~300V film across the vapor gap.