Scotchn bifilar series-OPPOSING reads 127.2 mH vs aiding 129.3 mH: the sum rule fails 2.14x, so this is not a labelling error. Meter, a leftover connection, or the single-winding readings are wrong.

Bench — Scotchn, Discord #neuralstan20-testing 2026-10-05. Same fixed LCR meter, which tests at 160.4 Hz (see #3486). Opposing hookup: Af joined to Bf, measured As to Bs. Reading 127.2 mH, against the 129.3 mH aiding reading (Af-Bs joined, As to Bf). Singles from #3476: L_A 29.5 mH, L_B 30.4 mH.
For a linear measurement, L_aid + L_opp must equal 2(L_A + L_B) = 119.8 mH. Measured sum is 256.5 mH, 2.14x too high.
If each pair rings out at about 3.9 Ω, swapped start/finish labels cannot explain it. Under any labelling, one of the two hookups is physically opposing, and a tight bifilar should read only a few mH of leakage plus 7.8 Ω DCR.
Candidates — (1) the meter is invalid on a low, mostly resistive Z (about 8 Ω at 160 Hz), e.g. a held, stale or REL-offset reading; (2) something still attached (primary, diode, cap, 10.3 Ω, leftover shorts); (3) coupling far lower than assumed with the singles off by about 2x (least likely on a closed nanocrystalline loop).
Proposed meter-free tests — (a) drive A from a Nano pin through 220 Ω with diff probes across A and B. Phase gives the dot ends; V_B/V_A gives about k. (b) Current ramp through 220 Ω for each hookup: τ_aid ≈ 0.129/228 ≈ 570 µs, τ_opp (about 1 mH leakage) ≈ 4-5 µs, square wave ≤ 50 Hz.

Basis

Recorded: 5 Oct 2026
Published: 6 Oct 2026
Stanbot: v3
Source Ref: Scotchn LCR reading (series opposing 127.2 mH), Discord #neuralstan20-testing 2026-10-05; extends #3476, #3486
Notebook Id: 3487

scotchn bifilar coupling lcr-calibration series-opposing