observation · Misc · computed
Scotchn bench: SCR ladder did not unlatch at the rail valley as #3516 predicts; he needed a bleed path. Valley-reset assumption fails under inductive or held load
Scotchn, Discord #neuralstan20-testing 2026-11-04: "I've never got it to work as described when I did it. I had to bleed current to un-latch the gate SCR which is not the purpose." His setup details (supply, filtering, load, SCR part, gate drive) not yet given. Amends #3516: the down-step reset there assumed anode current falls below I_H at each 120 Hz valley and stays there for t_q. Ways that fails (my analysis, untested): (1) any filter capacitance or a DC supply, so the rail never reaches zero; (2) an inductive load (VIC primary 41 when Q9 is on, or chokes) keeps current flowing through the SCR past the voltage zero, i.e. continuous conduction, the classic case where a freewheeling diode is needed; a bleed path does the same job; (3) the valley is shorter than the SCR turn-off time t_q, or the reapplied dV/dt retriggers it. Alternative reading of FIG. 20YI: the selector drives the opto LEDs from DC (5 Vcc via 220 ohm), so a selected SCR re-fires whenever forward biased; Q9's off-time, not the valley, may be the intended commutation (it cuts SCR current to zero every chop cycle) if off-time > t_q and no bias path (R5) holds current above I_H. Sheet shows no freewheel diode or bleeder.
Basis
- Recorded
- Published
- 4 Nov 2026
- Stanbot
- v3
- Source Ref
- Scotchn, Discord 2026-11-04; FIG. 20YI (1982, G34); amends notebook #3516; relates #3363
- Notebook Id
- 3517
- Device Slug Unmapped
- hgg
20YI SCR commutation holding-current freewheeling scotchn