Lab

PD Sensor Signal Chain Lab

A sensor with good saturation headroom is only half the story. What actually reaches your oscilloscope depends on the winding's frequency response and the electronics sitting between it and the scope — explored here for a synthetic 50 Hz + partial-discharge pulse, continuing where the core-geometry trade-off in Flux Concentrator Lab leaves off.

A trade-off you don't see from geometry alone

A published R-L bypass network strongly rejects the 50 Hz operating current that would otherwise swamp the measurement — the chart below reproduces its published 73 dB of attenuation at 50 Hz exactly. But the same filter also reshapes the much faster PD pulse it's meant to pass through: toggle it below and watch the pulse's ringing roughly double. Good 50 Hz rejection and a clean PD waveform are not automatically the same goal.

Output filter (Fritsch R-L bypass)

Secondary turns

PD pulse rise time

50 Hz operating current

Toggle the filter or pick a value Drag the view to rotate, scroll to zoom

PD pulse at the scope

0.622 V

peak, rise 0.71 ns

Overshoot/ringing: 3.7%

Linear, illustrative-material model — not a validated prediction.

PD peak at scope

0.622 V

Rise time

0.71 ns

Overshoot / ringing

3.7 %

50 Hz attenuation (filter)

-73.4 dB

Frequency response, Z_T(f)

The sensor's own transfer impedance (Fritsch & Wolter, IEEE TIM 2022) rises from 50 Hz up to a plateau; the output filter (orange) suppresses the low end further while converging to the same plateau at high frequency — the two always meet, by construction.

Plain 50 Ω burden With R-L bypass

Reconstructed PD pulse, v_out(t)

A double-exponential PD current pulse through the whole chain, bare vs. with the output filter. Redraws when you fire a pulse with the current controls.

Bare With filter

What the real papers actually measured

The 50 Hz attenuation row below is the published Fritsch & Wolter (POWERCON 2022) design example, reproduced exactly by the engine above. The THD-threshold rows are Zimmer et al. (2026) Table 3 — experimental data, not computed by this tool.

Gap / geometry	THD = 1% threshold	THD = 5% threshold
HFCT, 0.0 mm gap	16 A	23 A
HFCT, 1.1 mm gap	184 A	219 A
HFCT, 6.5 mm gap	384 A	440 A
MFCS (half-core)	>450 A — threshold not reached within the tested range (censored, not a measured value)

Diagnostic only, not the MFCS: at 450 A on the conventional 1.1 mm-gap HFCT geometry (the one geometry in this data set with a defensible 1D reading), the 1D model estimates B_peak ≈ 446.5 mT. No B_sat is asserted — this is reported, never compared to a verdict.

How this is built, and its real limits: this is a linear, unsaturated small-signal model — it cannot produce harmonic distortion by construction, which is why THD is never one of the numbers above. The transfer-impedance model and the output filter are combined as two cascaded linear blocks (multiplying their transfer functions), not solved as one unified circuit with the bypass as a third parallel branch — a reasonable first approximation, not a lab-accurate voltage prediction. The material constant (μ_r = 800) is a Fair-Rite 43 manufacturer reference value used for validating the conventional closed-HFCT model, not a measured MFCS material, and the magnetic-circuit diagnostic above uses the conventional small-gap HFCT geometry, never the half-ring MFCS — per Flux Concentrator Lab's own rule, the open MFCS core must not be treated as an equivalent air gap. Neither model includes winding self-capacitance, so the combined response plateaus rather than resonates at high frequency.

Explore: Flux Concentrator Core Geometry Lab

The upstream half of this story — how much core material a PD sensor actually needs, and the sensitivity it trades away by opening the magnetic path.