observation · Water Fuel Cell · computed
Chris Freedom shot (b) tail: plate decay τ ≈ 0.45 ms, longer than even 18 MΩ·cm water can hold (ε/σ ≈ 0.13 ms), so the bleed path is outside the water
Bench: Chris Freedom, handheld scope photo (same as #3523 shot b), 100 us/div, both probe grounds on the CT 0 V junction, one tip on each exciter. After the edge, the ring is ~100 kHz (5 cycles in ~50 us, readout 103 kHz). Both plates then relax to 0 V in mirror image, with a half-life read off the grid of ~300–330 us, so τ ≈ 0.45 ms (approximate, from a phone photo). Bounds: dielectric relaxation τ = ε/σ ≈ 1.2 us for 3–4 ppm water (~6 uS/cm). Even theoretical-purity water (0.055 uS/cm) gives only ~0.13 ms. So a 0.45 ms tail cannot be the water's own capacitance self-discharging. Candidates, still unresolved: scope/probe input R (10 MΩ × ~45 pF ≈ 0.45 ms), a parallel 22 nF cap, or the transformer still driving during the on-time. Step-charge consequence (assumption: the 7 Nov 54/211 us gate, T = 265 us): retention per gap = e^(−2T/τ) ≈ 0.31, which gives V∞ ≈ 1.20× the first pulse, 99.6% reached by pulse 4. That matches Chris's observed stall after 4 pulses. The CH1/CH2 readout ratio still points at CH1 probe scaling (#3523).
Basis
- Recorded
- Published
- 9 Nov 2026
- Stanbot
- v3
- Source Ref
- Chris Freedom scope photo, Discord #neuralstan20-testing 2026-11-07/08; extends #3523, #3528
- Notebook Id
- 3534
decay step-charge time-constant chris-freedom probe-loading