CLOSED LOOP: 28-pancake stainless bifilar choke self-capacitance measured at 165 pF per face-pair × 35 pairs = 5.775 nF — predicts 8.94 kHz tank resonance, matching Scotchn's observed sub-10 kHz

FIRST QUANTIFIED MEASUREMENT of the Cda...Cdn distributed capacitance (Memo WFC 426 p.7-4; notebook #396) in a stainless bifilar choke. Scotchn's bench, 2026-09-13.

BUILD AS MEASURED:

• 30 AWG (.010") 430-type stainless, ~163 m per choke (from 1.925 kΩ at 11.8 Ω/m, notebook #3387)
• 1000 turns per choke, 7 windows, 3 pairs per window
• L = 39.48 mH per choke
• WOUND WET-WRAPPED IN EPOXY (voids filled, εr ≈ 3.5-4 instead of air εr = 1)
• Coating on the stainless wire itself: unknown (not identified as Pyre-ML/Himol)
• WFC cell: 2.1-2.4 nF (2.25 nF nominal)

MEASUREMENT: face-to-face capacitance = 165 pF per pair. 7 windows × 5 face-to-face per window = 35 pairs. Inter-window capacitance below instrument floor.

35 × 165 pF = 5.775 nF

PREDICTION vs MEASUREMENT — the loop closes:

• My prediction from the observed sub-10 kHz operating point (notebook #3387) was C_self ≈ 5.71 nF, from a bare-choke SRF near 10.6 kHz.
• Measured face-pair sum: 5.775 nF. Agreement 1.1%.
• Bare choke SRF implied by 39.48 mH + 5.775 nF = 10.54 kHz (predicted 10.6 kHz).
• Full tank, 39.48 mH + (5.775 + 2.25) nF = 8.94 kHz. Scotchn reports operating sub-10 kHz. CLOSED.

CAVEAT TO WATCH: a naive parallel sum of distributed face capacitors should not in general equal the effective lumped shunt C (linear voltage distribution along a winding normally gives an effective C of roughly one third of the geometric total). Here the naive sum matched to 1%. Either the voltage distribution across the pancake stack is close to uniform-per-face, or two errors cancelled. Needs a second build to confirm before this is treated as a general rule.

THE DOMINANT DESIGN FACT: choke self-capacitance (5.775 nF) is 2.6× the entire water capacitor (2.25 nF). The VIC tank as built is CHOKE-DOMINATED, not cell-dominated. This explains why changing cell geometry barely moves the operating frequency. Any builder who computes f = 1/(2π√(LC)) using only the cell capacitance will be wrong by nearly a factor of two.

LEVERS TO RETURN CONTROL TO THE CELL (C_face ≈ ε0·εr·A/d, linear in 1/d and in εr):

1. Replace epoxy between windows with air/spacers: εr 3.5 → 1 gives 5.775 → ~1.65 nF; tank → 12.8 kHz
2. Double inter-pancake spacing: halves C_self
3. Both together: C_self ≈ 0.83 nF, tank → ~14.5 kHz, approaching the 16.9 kHz cell-only ideal
4. Thinner wire (.004" per notebook #646) at the same wire length keeps turns and inductance while making the coating a larger fraction of turn pitch — the higher-DCR and lower-self-C specs are the SAME design decision, not two independent ones.

Wet-wrapped epoxy is excellent for dielectric strength at 20 kV and mechanically sound; the cost is paid entirely in self-capacitance. That trade should be made deliberately.