8XA SCR commutates on the unfiltered 100 Hz rail, not on the tank

Analysis of the 8XA-style variac supply as drawn (variac -> full-wave bridge -> gate SCR -> blocking diode -> bifilar choke -> WFC cell, with an anti-parallel diode across the SCR), prompted by Scotchn's bench work.

Key point: the SERIES blocking diode makes loop current unidirectional, so the L-C tank can never reverse current through the SCR anode. Natural (tank) commutation is therefore impossible by construction. Turn-off can only come from the anode current decaying below holding current I_H, which on this topology means the rectified rail itself falling toward zero. The unfiltered 100 Hz rectified sine IS the commutation mechanism, which is why the drawn schematic has no supply-side storage.

Numbers using Scotchn's bench VIC (L = 2 x 39.48 mH = 78.96 mH, C_cell = 2.2 nF, loop DCR = 2 x 2.5676 = 5.14 kohm), NZ 230 Vrms mains, bridge peak 325 V:

• Tank f0 = 12.08 kHz, Q = 1.17, zeta = 0.43 (underdamped).
• R_crit = 2*sqrt(L/C) = 11.98 kohm.
• With Stan's quoted 11.6 kohm per choke (23.2 kohm loop) the same tank is zeta = 1.94, hard overdamped: no ringing at all, consistent with the claim that matched choke resistance prevents coil-ringing during pulse off-time.
• Standing DC current at rail peak: 325 V / 5.14 kohm = 63 mA (above typical I_H 20-40 mA); 325 V / 23.2 kohm = 14 mA (below it).
• Commutation window on an unfiltered rail with the 5.14 kohm loop: current drops below a 25 mA I_H when instantaneous rail < 129 V, i.e. sin(theta) < 0.397, theta < 23.4 deg. That is ~26% of each 10 ms half-cycle = ~2.6 ms of zero-current time per 100 Hz pulse, vastly more than any SCR t_q (5-200 us).

Diagnostic consequence: on a genuinely unfiltered rail this circuit should self-commutate at 100 Hz even with the low-resistance bench chokes. Persistent latching therefore points to (a) an unaccounted filter/reservoir capacitance, (b) a continuous or over-wide gate drive re-firing during recovery, or (c) reapplied dV/dt exceeding spec without a snubber - not to a mistuned tank.

Role of the anti-parallel diode across the SCR: it is a freewheel path for the choke's field collapse when the rail drops, not a commutation aid. It clamps the SCR to roughly -0.7 V only, which is weak reverse bias for t_q purposes. Consistent with Scotchn's observed ~375 V flyback vs ~200 V driven pulses.

Caveat: the damping coincidence at ~12 kohm is specific to the 39 mH bench chokes. With the VIC5 Estate chokes (reported >19 H each) R_crit rises to ~186 kohm and 23.2 kohm gives zeta ~0.13 - strongly underdamped. Bench-scale finding, not a universal law.