Steam Resonator design session v5: reasoning

Ran v5's sweep pushing the fundamental up into 1–30 MHz on the theory that a real cycle finally starts loading the Debye tail, while trying to keep amp flow restricted with low-resistance chokes (lean on ωL not R) and a small-C_geo, wide-gap, short-cell geometry, plus crushed conductivity (ultrapure, near-freezing water) so τ_bulk stays inside the on-time. Score 0.521, 24.63 kg/hr steam — the best yield number I've logged — but the heat budget gives it away: 0.0% non-ohmic. Of the 17.73 kW landing in the cavity, only 4.5 W is dielectric relaxation (Stan's "particle impact" mechanism); 8865 W is double-layer cycling and another 8865 W is straight ohmic leakage — 20.5 A leaking through a 1Ω pickup + 1Ω choke that were supposed to be restricting amp flow, not conducting it. This is a kettle, not a resonator. The resistive pickup/choke pair isn't doing its job at this operating point; the field is establishing to 91% in the 6 ns bulk relaxation time, so most of each phase window is spent with a fully-formed field sitting there driving conduction current rather than doing dielectric work.

The frequency placement confirms it: 30 MHz is nearly three decades below water's ~19 GHz room-temp Debye peak (56.9 GHz reported here reflects the ultrapure/near-freezing shift, still nowhere close). At ε''=2.66e-2, we're sitting at 5.27e-2% of peak loss — negligible dielectric coupling no matter how good the field establishment looks. Raising frequency toward the Debye tail helped ε'' a little, but conduction losses scaled right along with it because I let leakage current explode to hit yield.

Plan for next session: fix leakage current as the constrained variable, not a free parameter — cap it well below the amp-restriction target instead of letting the optimizer chase yield through it. Push fundamental frequency further, toward the 1–19 GHz range if the driver topology allows, since that's where ε'' actually rises. Reduce dead-time creep that's letting the field over-establish during on-time — target τ_bulk fraction under 50%, not 91%. Re-score with a heat-budget floor requiring non-ohmic fraction above some threshold (30%+) so the sweep can't converge on kettle solutions again.