VIC design session v62: reasoning

Lab notebook — VIC resonant sweep, 144 candidates, L(N) now tracking turns/gauge.

The headline is a contradiction I have to be honest about: the scorer keeps handing me the 0.1mH high-Q reference chokes because that's where coil Q and cell voltage peak, yet every one of my top five sits at f_r = 303.9kHz — a full violation of the 1-100kHz practical band. The intended win of this session, pushing per-choke L into 5-10mH ferrite to seat resonance at 40-60kHz, produced NO top candidates. That tells me the CGDE film-window reward is dominating the score and drowning out the frequency penalty. My objective function is mis-weighted, not my physics.

What the sweep genuinely shows: the trade is Q vs current restriction. Dropping wire gauge and adding turns (100→600T, AWG18→26) trades coil Q 709→88 and cell voltage 314→249V, all riding the same 1.37nF cell. Every candidate lights its 0.1mm vapor film at 99-125 Td — squarely inside the 3-6eV cold-chemistry window — so on discharge chemistry they're indistinguishable; frequency is the only discriminator, and they all fail it.

So I adopt none of these five for the bench. Provisionally I flag #1 (highest Q, 125 Td, cleanest film onset) as the physics reference only, not a buildable point.

Next session: (1) re-weight the score so the band violation is a hard gate before CGDE reward, (2) re-run the 5-10mH ferrite arm alone and confirm L truly tracks N² there, (3) if 300kHz keeps winning on chemistry, ask whether my "practical band" ceiling is the real constraint or an inherited assumption. Fix the scorer first.