observation · Voltage Intensifier Circuit · computed
Scotchn 10.3 Ω diode shunt: ms-scale positive CH1 tail (~1.8 V peak, ~0.9 V at 1.1 ms) is NOT probe common-mode error — shorted-lead test flat (277 mVpp noise). Charge budget still rules out current through the cap; the tail must be magnetic pickup in the
- Bench — Scotchn, Discord #neuralstan20-testing 2026-09-30. 10.3 Ω cement (wirewound-style) resistor between the diode cathode and AF; CH1 100X differential, + on the cathode side, − on the AF side (positive = forward diode current; 1 V ≈ 97 mA). Pot 1 kΩ, C_t 3.97 nF, CH2 across cap 47.8 Vpp. 200 µs/div, 16× average, 500 mV/div.
- Across the shunt — pulse-synchronous steps build the envelope to ~1.8 V by the end of the burst, then a slow decay (~0.9 V at 1.1 ms). Literal reading is ~175 mA peak and ~100 µC per burst, versus ≤ ~190 nC that can pass through C_t (3.97 nF × 48 V). That is ~500× too much.
- Shorted-lead test (both CH1 tips on the cathode end, same settings): flat, Vpp 276.93 mV of noise, no pulses, no tail. So common-mode feedthrough is NOT the cause.
- Remaining explanations — (1) EMF induced by the core's field in the loop formed by the probe leads spanning the resistor (resistor sits right beside the winding on the core); (2) a real current in a loop that bypasses C_t, i.e. an unaccounted DC path (e.g. the passive CH3 probe's ground clip, or another earth connection on the secondary).
- Discriminating test — jumper across the 10.3 Ω with the probe tips left in place. A tail that persists means pickup; a tail that vanishes means real current, and then trace the DC path.
Basis
- Recorded
- Published
- 30 Sep 2026
- Stanbot
- v3
- Source Ref
- Scotchn Rigol captures + bench photo, Discord #neuralstan20-testing 2026-09-30; extends #3454
- Notebook Id
- 3455
scotchn diode-shunt probe-artifact charge-budget