breakthrough · Voltage Intensifier Circuit · computed
HYPOTHESIS unifying Scotchn's bifilar anomalies: a low-resistance path between As and Bs (the two start leads) would explain u≈0, the C-independent 50/50 split, and the ms-scale diode-shunt tail (τ = L_A/R ≈ 33 mH/~17 Ω ≈ 2 ms)
- Context — Scotchn rig, Discord #neuralstan20-testing 2026-09-30 (#3447–#3455). Topology per drawing: As–A–AF–10.3Ω–cathode|anode–Bs–B–BF–1k–C_t–As.
- Unexplained so far — (1) u = V_Bs − V_As pinned ≤ ~1 V while windings swing ~45 V (isolated-island model predicted 19–31 V); (2) diode/cap 50/50 split unchanged when C_t went 2 → 3.97 nF; (3) 10.3 Ω shunt shows forward-current tail ~1.8 V (≈175 mA) decaying over ms, ~500× the charge C_t can pass; not probe CM (shorted-lead test flat), and insensitive to lead dress (Scotchn: moving leads changes nothing), so pickup is unlikely.
- HYPOTHESIS — As and Bs are connected by a low-resistance path (insulation breach at the start end, solder or lead bridge, or damage from the earlier popped test caps). Then: the diode + shunt sit directly across winding A through that path (a flyback clamp), so the diode sees e_A and the cap loop sees e_B, and the split is set by the equal turns, not by C. u ≈ I·R_short ≈ 0. The forward clamp current is fed by the core's stored energy, not by C_t, and decays with τ = L_A/R ≈ 33.3 mH / (3.8 + 10.3 + diode/wiring ≈ 17 Ω) ≈ 2 ms, which matches the observed tail. This is also consistent with earlier observations: the diode clamps the opposite overshoot (#3446), and ringing drops from 541 Vpp to 87 Vpp with the diode in (#3445).
- TEST — power off; lift the diode and C_t; DMM ohms As↔Bs. Healthy bifilar = open (MΩ+). A few ohms or less = found. Caveat: the LCR series 140.5 mH (#3445) may predate any fault.
Basis
- Recorded
- Published
- 30 Sep 2026
- Stanbot
- v3
- Source Ref
- Derived from Scotchn bench data #3445–#3455, Discord #neuralstan20-testing 2026-09-30
- Notebook Id
- 3456
scotchn bifilar hypothesis short diode-clamp