established · computed
Switch topology follows the rail: high-side NPN + opto for HV, low-side FET for 12-24V
Stan's drive stages split by rail voltage, and the split is deliberate:
HIGH-VOLTAGE RAILS (variac 0-120VAC, alternator step-up, 650VDC per the 2N6676 in the specs): HIGH-SIDE NPN power transistor (2N6676-class, 650V) driven through an H11D1 optocoupler with ~100Ω 10W base network fed off the source. Why high-side + opto, against convention: (1) GALVANIC ISOLATION — the pulse logic never shares a ground with the HV loop, so inductive kickback and ground bounce can't reach the timing circuits; (2) SAFETY — with the switch on the high side, the primary/load sits at ground potential when OFF: probing is safer and a fault de-energizes to ground instead of sitting live; (3) ERA-CORRECT DEVICE PHYSICS — 650V power BJTs were the robust HV switch of the day (high dV/dt tolerance, no gate-oxide fragility), and an opto-darlington can source base current from the collector rail through the 100Ω without the bootstrapped gate supply a high-side N-MOSFET would demand.
LOW-VOLTAGE RAILS (12-24V into the VIC5 core for the injectors etc.): LOW-SIDE switching, ground-referenced drive — no isolation needed at that potential, and a ground-referenced gate gives the fastest, sharpest edges (which the polarization process lives on).
DESIGN RULE going forward: when a design's drive_peak_v climbs past bench-safe low voltage (~24V), the driver spec should follow Stan's arrangement — high-side NPN + opto isolation, switch rated ≥ rail with margin — not just a low-side N-FET with a bigger V_DS number.
Basis
- Confidence
- 0.90
- Published
- 19 Aug 2026
- Origin
- StanBot lab notebook
- Component
- switching
- Source Ref
- Chris bench notes 2026-07-27; Stan's drive specs (2N6676 650V, H11D1 opto, 100Ω 10W); VIC5 injector core docs
- Recorded At
- 2026-07-27 18:12:10.883118
- Source Channel
- neuralstan20-testing
switching driver high-side opto-isolation 2N6676 H11D1 safety