WFC design session v8: reasoning

Lab notebook — resonant cell sweep, 108 candidates

Ran the parameter sweep with water treated as dielectric (εr = 78.54 at 20°C, per WO8912704A1 and #4,936,961) and the leakage resistance transformed to its series equivalent at resonance. The trend across the top five is clear: as annular gap and cell capacitance shrink slightly, f_r climbs from ~92.5kHz to ~98.3kHz, tank Q rises from 58.8 to 62.4, and cell voltage climbs from 705V to 748V — all while series current stays pinned near 605-606mA, comfortably under my 1A ceiling. The choke pair (2×1mH, 100t AWG24) is unchanged across all candidates, so the geometry of the water capacitor is really doing the tuning here, not the coil. This confirms the water-as-capacitor model: tighter rod/tube spacing (15.9mm rod in 22.3mm tube, per WO9207861A1-style geometry) lowers C, pushes f_r up, and buys Q without punishing current draw — the trade-off is milder than I expected. There's no sharp knee where higher Q forces current past the limit within this cluster.

I'm adopting #1: 15.9mm rod / 22.3mm tube / 101.6mm cell, 1.31nF, f_r = 98.33kHz, Q=62.4, 748V from 12V drive at 605.7mA. It's the top score, sits well inside the operating band, and gives the best voltage multiplication of the batch without added current risk.

Next session: sweep tube wall thickness and cell length independently (currently coupled), test 3-cell series/parallel stacking at this geometry, and vary choke turns count to see if Q gains can be had without touching cell geometry.