VIC design session v68: reasoning

The sweep didn't do what I aimed it to. My intent was to lift per-choke L into the 0.8–1.6mH heavy-gauge band to drag f_r under 100kHz, but every reported candidate still sits at L = 0.0001H (0.1mH) and f_r nailed to 303.89kHz. The turns column climbs (200→600) while inductance never moves — that's a model bug, not physics. Two hundred turns on this former cannot be 0.1mH; the L-from-geometry path clearly isn't being fed the swept turns. So I treat the frequencies here as untrustworthy and the "adoption" as provisional.

What the numbers DO show cleanly is the Q-versus-wire trade. Holding L fixed, adding turns/wire only bleeds coil Q: 354→322→236→158→118 from #1 to #5, dragging tank Q 25.2→22.1 and cell voltage 303→265V with it. More copper here buys nothing but loss — the 120 Td cold-chemistry window is held by the top three (121/120/117 Td), and they're also the highest-Q, lowest-wire builds. So within this (broken) run, #2 is the honest pick: AWG18/220t matches my adopted best, 120 Td dead-center, coil Q 322, only a hair behind #1's 121 Td while using less wire and staying consistent with prior sessions.

But all five carry the >100kHz band violation, so I'm adopting #2 only as bookkeeping, not as buildable.

Next session: fix the inductance calculation so swept turns actually raise L, re-run the 0.8–1.6mH band, and confirm f_r drops toward 40–70kHz while watching whether coil Q survives the heavier gauge and the Td window holds. Until L moves, the frequency axis is meaningless.