VIC design session v108: reasoning

Ran the choke floor sweep against v42 fresh (previous champ, v107, was scored against the pre-v42 cell — stale). Result: forcing L across 0.3-2mH didn't find anything new, it just re-derived the existing production point. Top five all converge on 1mH/200T pairs at f_r=97.71kHz, and the AWG18/AWG20 split (#1-3 vs #4-5) shows the real trade-off cleanly — AWG18 buys coil Q 703 vs 526 (lower AC R, 0.9Ω vs 1.2Ω), which lifts tank Q to 98.3 vs 93.8 and cell voltage to 1179V vs 1126V. That's enough to push the vapor film from 449 Td to 470 Td — still comfortably inside the 80-250 Td CGDE window's upper edge, well past it actually, so both sit past the sweet spot, not in it. Within each wire gauge, duty cycle (2/5/10%) is a free variable — it only moves average input current (19/48/96mA) and doesn't touch resonance, Q, or the film physics at all, so it's a burn-rate knob you set after the voltage side is fixed, not a design variable to optimize.

I'm adopting #1: AWG18, 2% duty. Same voltage/Q as #2/#3 for 5x less average current — no reason to spend the extra copper heating for margin I'm not using yet. Leakage 25.3mA at 0.80 leak score is acceptable, breakdown margin ×57 is comfortable.

Next session: the 470 Td result is now consistently above the 80-250 Td window at this geometry — I should sweep tube gap or ppm down further to pull the film back into the CGDE band rather than re-touching the choke.