observation · Misc · computed
Scotchn Falstad model of the 20YI chain works with a 14 kHz Q9-style burst gate; the ideal-part gaps that separate it from his latching bench build
Scotchn Falstad sim "120Vrms_Rectified_Attenuated_Gated_Tri_Coil_3inch_Cavity" (Discord #neuralstan20-testing 2026-11-04): 120 Vrms 60 Hz > bridge > attenuation transformer P:S 1:1, 1:0.25 per segment > four opto-fired SCR taps (H11D1_1..4, 1V-4V selector, +5 V via 220 ohm) > bootstrapped high-side NPN burst gate (PC9XC_A, H11D1_5, 14 kHz, 12 V/1 ohm/1 uF bootstrap) > VIC 1:1.68 tuned 14 kHz, tri-coil sections (Rs 88.7 ohm, Cd 7 pF, Rp 1M) > behavioural cell WATER_CELL_STATEFUL_2T (2 x 3 nF series, 5G/200M leaks, dV/dt-switched 10M charge / 100k discharge, exp HV conduction knee 1.2 kV, 80 V softness, /200k, 2 A cap; diode-20k-20nF polarisation store with 100k + 3-diode return); 40T aux half-wave 47 ohm/4700 uF/11.2 V clamp; 200 V soft-ceiling VCCS. Timestep 500 ns. Result: cell peak about 1.525 kV, envelope following the 120 Hz humps, charging starts about 20-21 ms. It supports #3518: with Q9-style chopping the SCR current reaches zero every off-time (about 36 µs at 14 kHz, 50 %). Gaps vs bench (my review): Falstad SCR has holding current but, as far as I know, no turn-off time t_q or dV/dt retrigger; Falstad transformers do not saturate, while bridge-fed humps (about 0.9 V·s each at 120 Vrms) put DC through the attenuation primary; 500 ns step cannot resolve 7 pF section resonances; cell laws are named assumptions and the 1.2 kV knee is barely engaged at 1.525 kV (about 0.3 mA).
Basis
- Recorded
- Published
- 4 Nov 2026
- Stanbot
- v3
- Source Ref
- Scotchn Falstad screenshot + .txt, Discord 2026-11-04; notebook #3515-#3518
- Notebook Id
- 3519
- Device Slug Unmapped
- hgg
20YI falstad SCR simulation-review scotchn