VIC design session v99: reasoning

This sweep tried to reach the 80-250 Td window from the inductance side — hold peakV modest, ladder the choke around 581µH to drag f_r off the water-branch-limited point and settle Q at 4-6. It did not do what I asked. Every top candidate converged on the SAME operating point: 1.1mH chokes, f_r 98.5kHz, tank Q 8.8, ×8.8 magnification. The optimizer preferred the higher-inductance rung because it delivers more cell voltage, and the scorer rewards that even though I wanted lower Q. So the "land Q in 4-6" objective failed — Q came in at 8.8, and the vapor film lights at 350-421 Td, which is ABOVE my 80-250 Td / 3-6 eV cold-chemistry target. These are overshooting into full avalanche, not the CGDE sweet spot. The identical 0.421 scores across #1-5 are just duty-cycle and drive-voltage permutations of one geometry; the sweep found a plateau, not a peak.

The honest trade-off: pushing Q up from the inductance side buys V_cell but blows past the Td window, and the water branch (154.6Ω series R, 89.8% displacement) isn't the limiter here — the choke reactance is. More Q ≠ better once the film over-lights.

I'm provisionally adopting #1 (100V drive, 3% duty, 19mA average) — lowest input current, cleanest thermal case, and its 350 Td is the closest of the batch to the window ceiling.

Next session: DROP inductance below 581µH and cut peakV to 60-90V to pull the film down toward 200 Td, and re-weight the scorer to PENALIZE Td above 250 so it stops chasing voltage past the cold-chemistry edge.