Drive voltage + duty cycle are now design levers (drive-duty-v3)

Chris opened two levers on the desk (2026-07-27) and I built them into the design loop (evaluate() → drive-duty-v3):

1. DRIVE VOLTAGE (peakV). The loop was hardcoded to a 12V continuous drive. V_cell = Q·V_drive is linear, so raising it is pure territory gain: v19's Q=101 cell goes 1215V → 12.1kV at 120V drive → 40.5kV at 400V. That is finally my real 20kV+ range across the array.

2. DUTY CYCLE. The old model assumed CONTINUOUS drive and reported the full circulating current I=V/R, then hard-capped it <1A. That was wrong for how the cell is actually run. Behind the blocking diode the tank rings on stored energy and the supply only tops it up during on-time, so the AVERAGE input current the bench meter reads is ≈ duty·(V/R). That is why Chris holds <200mA at 120V. I now report BOTH: i_circulating (CW loop current — rate the wire/switch for it) and i_input_avg (supply draw). The <1A rule is now a SOFT preference in the score (full credit ≤200mA, half at 1A), not a hard violation.

Revisit of v19 under the new model: the duty knob alone lifts its score 0.772 → 0.845 by pulling average draw to ≤200mA while HOLDING (or raising) cell voltage. 120V @ 2% duty = 12.1kV cell at 196mA. 400V @ 0.6% = 40.5kV at 196mA.

HONEST finding on the plasma term: v19's cell already sits at ~484 Td across the 0.1mm vapor film — ABOVE the 80-250 Td cold-chemistry (CGDE) window. Raising drive makes it WORSE, not better: 120V → 4843 Td (streamer), 400V → 16000+ Td (arc, model out of range). So higher voltage buys reach and, with low duty, low current — but it does NOT buy the cold-plasma window. The window wants LOWER cell voltage, i.e. lower tank Q (~20-30, giving a few-hundred-volt cell). I confirmed I canNOT get there by shrinking the choke — smaller L shoves f_r out of the 1-100kHz band (same lockout the VIC sessions hit at 0.1mH/304kHz). It needs a deliberately lossier tank at HELD f_r: a resistive choke leg or higher-ppm water to drop Q without moving resonance.

Next session: sweep a resistive-leg / higher-loss tank that lowers Q to ~20-30 while pinning f_r in-band, and find the drive voltage that lands the 0.1mm film at ~150 Td (mid cold-window) with duty set for <200mA. That's the config that could finally score full plasma credit AND full current credit together.