Minimum pshr (dB) for a candidate frame to count as sub-harmonic-CLEAN.
Evidence-set on the #116 corpus, margin on both sides: the four live false positives
(musical overtones at 1 869.1 / 832.0 / 263.7 / 228.5 Hz) read PSHR between −10.2 and
+8.8 dB across their windows — never a 5-frame sub-clean stretch, and the single
closest approach (+8.8) is isolated between frames at ≈ −1 and −10 — while the
committed #106 injected ramp reads ≥ +12.2 dB from its second frame on (its frame 0
sits at +9.4, costing exactly one frame of confirm latency, pinned in the corpus
test). A static program line at f/2 of REAL feedback also clears this fast: the
runaway gains several dB per hop on anything static, so the deferral is transient.
Minimum pshr (dB) for a candidate frame to count as sub-harmonic-CLEAN. Evidence-set on the #116 corpus, margin on both sides: the four live false positives (musical overtones at 1 869.1 / 832.0 / 263.7 / 228.5 Hz) read PSHR between −10.2 and +8.8 dB across their windows — never a 5-frame sub-clean stretch, and the single closest approach (+8.8) is isolated between frames at ≈ −1 and −10 — while the committed #106 injected ramp reads ≥ +12.2 dB from its second frame on (its frame 0 sits at +9.4, costing exactly one frame of confirm latency, pinned in the corpus test). A static program line at f/2 of REAL feedback also clears this fast: the runaway gains several dB per hop on anything static, so the deferral is transient.