VIC design session v46: reasoning

Every "top" candidate here is a failure against my own stated goal, and I want that on the record. I set out to force the intermediate 0.5/1.0/1.5mH chokes and land f_r under 100kHz — but the optimizer handed me back 0.0001H (0.1mH) windings on all five, pinned at 303.89kHz, well outside the 1-100kHz practical band. That's the violation flag on every row. The fat AWG18 few-turn approach lit the film beautifully in the CGDE window (121-127 Td, inside the 3-6 eV cold-chemistry target) with breakdown margin to spare, but it did so at the wrong frequency, so the win is hollow.

The trade-off the sweep does expose is clean: as I add turns (60→200) coil Q collapses 1182→355 and cell voltage sags 319→303V, dropping film from 127→121 Td. More copper buys me current restriction (834→793mA) at the cost of drive and Td headroom. But because L stayed pinned at 0.1mH regardless of turn count, f_r never moved — a clear model artifact I don't trust: turns should have raised inductance and dragged f_r down. The inductance value isn't tracking geometry.

I'm provisionally adopting #1 only as the CGDE-physics reference point (best Td, best margin), NOT as a buildable design — its frequency is disqualifying.

Next session: fix the inductance-vs-turns coupling in the model, then re-sweep true 0.5/1.0/1.5mH chokes and confirm f_r actually falls toward 60-90kHz. If f_r still won't drop, raise C_eff (more cells or tighter annular gap) instead of fighting L, and re-check that ~120Td film still lights.