VIC design session v76: reasoning

Sweeping L from 0.0001 to 0.0025H did NOT honestly walk f_r down into the 40-70kHz band the way the topology-only intuition promised — and the reason is written plainly in the results: every top-five candidate sits pinned at 303.89kHz on the SAME 0.0001H choke. That's the tell. With C_eff locked at 1.37nF, f_r = 1/(2π√LC) only falls as √L, so reaching 60kHz needs ~25× more inductance — and when the optimizer took that L, cell voltage collapsed out of the 3-6 eV / 80-250 Td cold-chemistry window and the score cratered. So the higher-L candidates that DO honor the 1-100kHz band never make the leaderboard; the film goes dark. The frequency violation is real and I'm not hiding it.

The Q-vs-restriction trade shows cleanly across #1-#5: winding more turns (130→480) buys coil Q (546→148, inverted here by AC R growth) but tank Q slides 25.9→22.9 and cell voltage 311→275V, dragging the film from 124 down to 110 Td. More copper, less light. Restriction is fine — 813mA, displacement 96.5%, breakdown ×1000+.

I'm adopting #1: 130t / AWG18 / 0.0001H, tank Q 25.9, 311V, 124 Td — deepest into the CGDE window with the least wire. It carries the frequency violation, so it's a bench-only reference point, not a build.

Next session the honest lever is C, not L: raise C_eff (more parallel cells, tighter annulus, longer tubes) so a band-legal larger L can co-drop f_r toward 70kHz WITHOUT starving the film. Sweep C and L jointly, holding cell voltage ≥290V.