VIC design session v18: reasoning

Ran the low-L bracket (0.02-0.15mH) tonight expecting to pull f_r down into band from above, but the geometry backfired — every top candidate clustered at L=0.1mH and landed at 326.8kHz, nearly 5x over my 70kHz ceiling. Cutting L that far didn't help because C_eff is pinned at 1.19nF by the 9mm/13mm cell geometry, and f_r = 1/(2π√LC) just doesn't have room to fall into 40-70kHz without either much larger L or a much larger C_eff than this cell gives me. So this low-L branch is a dead end by construction, not by tuning — confirms the mid-L region (0.3-0.8mH range from prior sweeps) is structurally where the band lives for this cell size.

That said, the Randles data here is informative: displacement current is pinned at 96.7% across all five, confirming the double-layer capacitance (21.9µF) dominates and ionic current is genuinely minimal at 2ppm — consistent with my notebook's polarization thesis. Breakdown margin is enormous (380-414x) at this purity, so I'm not electrode-limited, I'm topology-limited.

I'm not adopting any of #1-5 — all violate the band constraint outright, and score differences between them are just turns/AWG trading coil Q (251 down to 137) for marginal voltage loss (314V to 289V), not worth resolving until frequency is fixed.

Next session: abandon this cell geometry for the low-L branch entirely; instead sweep larger rod/tube gap or longer cell length to push C_eff up 5-10x, which brings f_r into band without needing exotic inductor values.