WFC design session v30: reasoning

This sweep did not do what I asked it to. I aimed the electrode-gap ladder at the 80-250 Td cold-chemistry window, and every top candidate overshoots it — #1-3 light at 319 Td, #4-5 at 478 Td. All are townsend-avalanche across the 0.1mm film, above the 3-6 eV CGDE band I was scoring toward. The 4.80mm gap at 800V simply drops too much field across a 0.1mm vapor film. The scorer still crowned these at 0.512 because they resonate cleanly and keep breakdown margin high (×256-384) — the Td term isn't yet penalizing hard enough. That's a scoring-weight flaw to fix, not a real win.

The genuine trade-off is visible though: raising V_cell (800→1199V) slides Td the WRONG way (319→478) while doubling tank current (1.27→1.90A) and Faradaic leakage (1.3→2.0A). So brute voltage costs current AND pushes me further from the window. Tank Q sits fixed at 13.3 across all five (same 45.7kHz, same 1.1mH chokes) — Q wasn't actually a lever this round.

I'm adopting #1: lowest V_cell (800V), lowest duty (2%, 25mA average input), best breakdown margin. It's the least-bad seed, not a target hit.

Next session, the fix is geometric, not electrical: to reach 80-250 Td at ~800V I need a LARGER gap or a thinner assumed film — widen the 4.80mm gap toward 7-9mm and let Td fall into band, then re-check whether C_eff still resonates near 45kHz. Also reweight the scorer so out-of-window Td actually costs points. Note the caveat: this is model, not bench — the 0.1mm film thickness is an assumption I haven't measured.