VIC design session v103: reasoning

Lab note — VIC resonant sweep, 600 candidates, Randles + vapor-film ladder.

First, an honest failure of intent: I set out to raise L on AWG24 chokes to drag f_r from 94.87 kHz down into 55–75 kHz and land the film inside the 80–250 Td cold-chemistry window. The optimizer did neither. Every surviving top candidate sits at 85.70 kHz on AWG22 0.7 mH chokes, and the film lights at ~313–320 Td in a townsend-avalanche regime — above the CGDE window I was rewarding. The scorer clearly traded band-placement away to hold tank Q and breakdown margin. So the headline result is that with this cell (2.46 nF, 5874 Ω bulk) and a 120 V drive, I cannot simultaneously reach 55–75 kHz AND stay sub-breakdown AND drop into 80–250 Td — the levers fight each other.

The Q trade is visible in #1→#4: pushing turns 900→1050 barely moves tank Q (6.7→6.6) while coil Q collapses 52.8→45.2 and film Td drops 320→313. More copper buys almost no magnification, just resistive loss. Not worth it.

Faradaic leakage is the real problem: ~415 mA at peak, leak score 0.19, and τ_EDL is absurd (34,000 carrier cycles to charge) — the double layer never fully forms at 85 kHz. Duty only scales average input (21/43/85 mA), not the physics.

Adopting #1: lowest duty (2%, 21 mA average), highest coil Q, same 803 V cell voltage. Cleanest of an imperfect set.

Next session: drop drive frequency by increasing C (bigger cell / parallel tubes) rather than L, and reweight scoring to actually penalize >250 Td so the film lands in-window.