observation · Voltage Intensifier Circuit · computed
Scotchn midpoint-diode rig, drain and energy session: 4 V supply, FQP30N06L (60 V) avalanching at ~78 V spikes; ~0.17 mJ in per burst vs ~0.16 mJ cap swing
- Bench — Scotchn, Discord #neuralstan20-testing 2026-10-05 (Rigol photos, supply and controller photos). Extends #3477, #3480.
- Drive — Arduino Nano burst generator (isolated supply), F_sw 2930 Hz, 50 % square, P = 1 pulse per burst, G = 14322 µs gate/relaxation time (about 70 Hz PRF; G is adjusted by clock ticks to overlay the first leading edge). Supply 4.00 V, 0.003 A shown (~12 mW, ±~15 % at 1 mA resolution). Session counter 0.458 Wh / 4:37:50 spans mixed settings: not usable.
- Drain (CH4, 10x passive, after correcting a 1x/10x mis-set) — plateau ~20–25 V during flyback (4 V + ~20 V reflected, i.e. ~2 V/turn, ~880 V across the 440-turn secondary, consistent with ~+300 V cap plus ~−450 V across the diode); spike ~78 V at turn-off, ~55–60 V at each packet onset, ~2 MHz ring (primary leakage vs Coss + snubber). FET: FQP30N06L, 60 V rated, so the ~78 V peak is probably its avalanche clamp, not the natural spike. FET stone cold at 4 V. The spike scales with supply (~20x), so raising the supply needs a higher-Vds FET or a clamp below 60 V.
- Energy (rough) — in ≈ 12 mW / 70 Hz ≈ 0.17 mJ per burst; cap ½·3.97 nF·(300² − 100²) ≈ 0.16 mJ peak swing, then returned and dissipated (the cap ends near its starting negative bias). Same order of magnitude only; not an efficiency figure.
- Pending — shunt in the FET source, then ∫V·I per pulse vs cap ΔE from the same capture at 100 V/div; a 1 h counter run at fixed settings; LCR across A–B; 1 nF shift test; midpoint-R sweep; diode in the proper VIC position.
Basis
- Recorded
- Published
- 5 Oct 2026
- Stanbot
- v3
- Source Ref
- Scotchn bench captures and photos, Discord #neuralstan20-testing 2026-10-05
- Notebook Id
- 3481
scotchn mosfet avalanche energy-balance bifilar drain-spike