123 Main Street, New York, NY 10001

Submit Your BOM

Dielectric & Partial Discharge (PD) Measurement

Dielectric & Partial Discharge (PD) Measurement

← Back to: Test & Measurement / Instrumentation

Dielectric partial discharge (PD) measurement turns fast, wideband pulse sensing into repeatable, phase-stamped PRPD evidence—so PDIV/PDEV decisions and trends are based on calibrated signals, audited gating, and complete setup/log traceability. A good system balances sensitivity, overload recovery, and safety interlocks, making results comparable across ranges, bandwidths, and real-world EMI conditions.

H2-1 · What PD measurement is (and what it isn’t)

Partial discharge (PD) measurement characterizes impulsive discharge events that occur inside or on the surface of an insulating system without a complete breakdown. The measurement target is not a steady “leakage value” but a stream of short pulses whose timing, repetition rate, and phase distribution reveal the presence and severity trend of defects such as voids, sharp-field regions, and interface irregularities.

The most common engineering quantity is apparent charge (pC). It is an equivalent number derived from a calibrated injection and the transfer behavior of the measurement path. It supports repeatable comparison under the same setup (sensor, bandwidth, gain/range, coupling network, wiring/fixture), but it does not represent the microscopic charge distribution inside the defect. For credible trending, the measurement settings and calibration state must be tied to every reported pC value.

What the key PD terms are used for
  • PRPD (phase-resolved PD): bins detected events by phase (0–360°) and magnitude to expose patterns, noise coupling, and gating effectiveness.
  • PDIV: the excitation level where repeatable PD events first satisfy the detection rule (window length + gating + threshold).
  • PDEV: the level where PD activity falls below the same rule on the way down, often differing from PDIV due to hysteresis and conditioning.

Page scope: this page focuses on wideband excitation/sensing paths, front-end sampling AFEs, phase-synchronous readout, and safety interlocks that make high-voltage PD tests repeatable and evidence-driven.

PD measurement scope map: DUT, sensing paths, AFE, PRPD and safety interlocks Block diagram showing a dielectric test object driven by a high-voltage source, multiple coupling and sensor paths feeding a wideband AFE, digitizer and trigger, producing PRPD output, with a bottom safety interlock chain gating HV enable and discharge. PD Measurement Scope Wideband sensing + phase sync readout + safety interlocks HV Excitation AC / pulse / ramp Enable gated by interlock DUT / Test Cell Electrodes Void / sharp-field region Coupling & Sensors IEC 60270 Path Coupling C + measuring Z HFCT Ground-loop pulse pickup UHF Near-field / radiated pickup TEV Surface transient pickup Wideband Readout Chain Wideband AFE Protection · Gain/range · Anti-alias ADC + Trigger Event slices Phase Ref 0–360° stamping PRPD / Evidence Phase bins · magnitude bins · event rate Safety Interlocks Door / E-Stop Hardwired loop HV Enable Default de-energize Discharge Path Bleed + verify Residual V Safe-to-touch Evidence Logging Enable reason · trips · timestamps Interlocks gate HV excitation and enforce discharge verification before access.
Figure F1 — A scope map that ties DUT + sensing paths to wideband AFE, phase-stamped PRPD evidence, and the HV interlock chain.

H2-2 · System topology: excitation + coupling + sensing paths

A PD measurement setup is best understood as one excitation path (driving the DUT) plus one or more sensing paths that capture fast discharge pulses. Different paths “see” different signatures, and they trade off frequency span, noise susceptibility, and deployment repeatability. Multi-path capture is often the most reliable way to separate true PD from external impulsive interference.

Four common sensing topologies (how they differ in practice)
  • IEC 60270 coupling (Coupling C + measuring impedance) — standardized transfer path for charge-based reporting. Sensitive to wiring/fixture configuration; best when calibration discipline and repeatability are prioritized.
  • HFCT on ground return — captures fast current pulses on the return conductor. Fast to deploy, but strongly dependent on return path geometry and sensor position.
  • UHF near-field pickup — captures radiated components from discharge activity. Useful for spatial discrimination and interference separation, but affected by enclosure/cavity structure and placement.
  • TEV surface pickup — captures surface transient voltages caused by discharge-related fields. Convenient contact-style sensing, but applicability depends on DUT construction and surface coupling.

Wideband capture becomes necessary when pulses are very short, when external interference is impulsive, or when evidence quality depends on feature extraction (pulse width/shape/time-of-arrival) and cross-channel correlation. In those cases, the system is designed so that each sensing path has a controlled bandwidth and gain/range, and all channels are aligned to a shared time and phase reference before PRPD and gating decisions.

A practical multi-path architecture keeps configuration “audit-ready”: each channel reports its sensor type, bandwidth setting, gain/range state, and calibration tag so results remain comparable across repeats and across fixtures.

Four sensing paths in one PD test setup with synchronous sampling Diagram showing a DUT driven by HV excitation and four parallel sensing paths (IEC 60270 coupling, HFCT, UHF and TEV), each with a front-end block feeding a shared synchronous digitizer and trigger, producing PRPD and event log outputs. Four Sensing Paths (One Setup) Parallel capture → per-path front-end → synchronous digitizer → PRPD & gating DUT / Test Cell Excited by HV HV Source Enable + interlock gate Sensing paths IEC 60270 Coupling C + measuring Z HFCT Clamp on return path UHF Near-field pickup TEV Surface pickup Front-End Gain · filter Front-End Protection Front-End Wideband Front-End Band control Sync ADC Aligned clocks Trigger + event cuts PRPD Phase-stamped events Gating Log Keep / reject evidence Multi-path capture improves noise separation by correlation and preserves repeatability via per-channel configuration tags.
Figure F2 — Four parallel PD sensing paths converging into synchronous digitization for PRPD and evidence-based gating.

H2-3 · Front-end AFE design for PD pulses (wideband, high dynamic range)

A PD front-end must preserve short pulse evidence while surviving a harsh HV environment. The design target is not only “gain and bandwidth,” but event detectability: pulses must cross a stable trigger rule, remain comparable across repeats, and avoid missing events after large transients.

Input protection: survive without destroying pulse features
  • Energy limiting first: a controlled series element and a defined return path prevent high-energy interference from directly hitting the amplifier input.
  • Fast limiting with fast recovery: clamps should prevent overvoltage but must not create long tails; slow recovery creates a “blind window” that reduces event rate and distorts PRPD statistics.
  • ESD/surge routing: keep surge current away from sensitive reference nodes; protection that injects current into the measurement reference will look like PD.

Wideband amplification is a balance between noise density and usable bandwidth. A practical PD AFE uses gain/range states (programmable attenuation or multi-gain stages) so small events are not buried by noise, while large pulses do not saturate the chain. Differential signaling helps when the installation is noisy or the return path is uncertain; single-ended can be sufficient when coupling and grounding are tightly controlled.

Anti-alias conditioning must be designed as a measurement bandwidth contract. Over-filtering can round the leading edge, reduce peak amplitude, and blur pulse-width features—changing trigger behavior and making results non-comparable across settings. For repeatable evidence, the selected bandwidth and range state should always be recorded as part of the measurement configuration snapshot.

AFE acceptance checks (evidence-first)
  • Overload recovery is short enough to avoid masking subsequent events at the expected repetition rate.
  • Noise floor within the selected bandwidth supports a trigger threshold that meets the minimum detectable event goal.
  • Range states prevent frequent saturation while keeping small events above the noise and ADC quantization.
  • Bandwidth setting is controlled and logged so PRPD and pC trending remain comparable over time.
Wideband PD AFE block: protection, gain/range, amplification, anti-alias and ADC Block diagram of a PD wideband front-end showing input protection and limiting, programmable range, wideband amplification, anti-alias conditioning and ADC input, with callouts for overload recovery, noise floor and bandwidth control. Wideband AFE (PD Pulses) Preserve pulse evidence · avoid saturation · enforce bandwidth contract Sensor Input PD pulse + interference Protection Limit energy Fast clamp ESD / surge path Overload recovery Avoid blind window Range State Attenuation Gain steps Saturation control Wideband Amp Low noise density Stable baseline Fast settling Noise floor Sets trigger margin Anti-alias BW control Shape keep No over-filter ADC In Headroom No clipping Evidence goals Pulse shape kept Edge + width features Trigger stable Threshold not drifting No blind window Fast recovery Config logged BW + range tags Key trade-off: bandwidth improves pulse fidelity but increases integrated noise; range states protect dynamic range and evidence quality.
Figure F3 — A PD AFE is an evidence chain: protection and recovery protect event completeness, while bandwidth and noise floor govern detectability.

H2-4 · Sampling & timing: ADC choice, trigger, and synchronous phase reference

Sampling is where PD evidence becomes measurable: events must be detected consistently, cut into comparable waveforms, and stamped with an accurate phase angle (0–360°) for PRPD. The sampling plan must match the analog bandwidth contract: too slow blurs edges and time-of-arrival, while mismatched anti-aliasing introduces shape changes that shift trigger behavior.

Practical sizing rules (Fs, bandwidth, and event fidelity)
  • Choose a measurement bandwidth that preserves the features needed for separation (edge, width, energy), then enforce it with anti-alias filtering.
  • Select sampling rate so edges are represented by multiple samples and event timing is stable; insufficient Fs makes triggers “jittery” even when pulses are real.
  • Ensure ADC headroom across range states; clipping and recovery artifacts create false pulse features and corrupt PRPD density.

Triggering should be chosen as a noise-aware decision: a simple threshold trigger is fast but sensitive to baseline drift; an energy trigger is more robust to noise but depends on the integration window; a matched-filter trigger strengthens detection when a consistent pulse signature is expected, but must be guarded against impulsive interference that “looks similar.” For repeatability, trigger settings and dead-time rules should be logged with every run.

PRPD requires a stable phase reference derived from the excitation waveform. Events are time-stamped, mapped to phase, then accumulated into phase bins. Phase jitter smears the distribution and makes patterns harder to interpret; a clear PRPD plot is usually the fastest confirmation that phase locking and stamping are working.

Sync sampling and phase stamping pipeline for PRPD Diagram showing continuous ADC sampling, trigger decision, event slicing, phase reference extraction, phase stamping to 0–360 degrees, and PRPD point cloud accumulation, with callouts for timing accuracy and phase jitter blur. Sync Sampling → Phase Stamping → PRPD Continuous samples become phase-resolved events with comparable slices ADC sample stream Wideband waveform Trigger Noise-aware rule Threshold Energy window Matched filter Event slice Comparable windows TOA + features Phase reference 0–360° from excitation Locked phase Phase stamping time-stamp → phase angle 0° · 180° · 360° PRPD accumulation phase bins vs magnitude Phase jitter → PRPD blur Record trigger rules, bandwidth, sampling rate, and phase-lock status to keep PRPD evidence comparable across runs.
Figure F4 — Sampling and timing turn waveforms into phase-resolved PD evidence: trigger + slicing + phase stamping produce PRPD patterns.

H2-5 · Detection & phase-sensitive readout (pulse features, I/Q, and PRPD)

Detection converts the sampled waveform into a set of events that can be counted, compared, and phase-resolved. A robust detector identifies candidate pulses, measures stable features, and attaches a phase angle so the same PD behavior produces the same PRPD signature across repeat runs.

Practical PD pulse features (what to extract and why it matters)
  • Peak / envelope — fast magnitude proxy for triggering, ranking, and saturation checks.
  • Energy in a window — more noise-robust than peak when interference is impulsive.
  • Rise-time / width — shape cues used to separate PD-like pulses from external interference.
  • TOA (time of arrival) — enables cross-channel correlation and stable phase stamping.
  • Quality flags — clipping, baseline drift, or post-overload recovery markers protect auditability.

Phase-sensitive readout is used here only in the PD context: when a periodic or synchronized reference exists, integrating an event in a phase window, or projecting it onto I/Q components, can improve effective SNR and stabilize classification in high-noise environments. The goal is not continuous lock-in measurement, but event scoring that remains consistent across phase and across channel conditions.

PRPD is then built by accumulating events into phase bins (0–360°) and amplitude bins, producing a density map. Typical patterns help confirm whether the system is stable (clean phase lock, consistent trigger) and whether interference is being gated correctly, without expanding into material-specific failure physics.

Event extraction pipeline: waveform to detection, features, phase stamping and PRPD Pipeline diagram showing sampled waveform input, event detection, feature extraction, phase stamping to 0-360 degrees, and PRPD heatmap accumulation with labeled phase and amplitude bins. Event Extraction Pipeline Waveform → detection → features → phase → PRPD density Waveform ADC samples Detection Find events Trigger rule Dead-time Features Stable metrics Peak / energy Width / rise TOA + flags Phase 0–360° Phase bins PRPD density map Phase bins (x) · Amplitude bins (y) · Count density phase bins → amplitude bins Keep the pipeline auditable: store feature definitions, phase stamping method, and clipping/quality flags with every run.
Figure F5 — Events are extracted from waveforms, scored by features, stamped by phase bins, then accumulated into a PRPD density map.

H2-6 · Noise separation & gating (how to reject EMI without hiding PD)

Field PD measurement often fails for one reason: interference dominates the event stream. Effective gating rejects EMI without deleting real PD by using multiple, explainable criteria and leaving an audit trail of what was removed and why.

Common interference types (what they look like to a detector)
  • Repeatable narrowband — periodic hum or carriers that leak into the band and create false triggers.
  • Random impulsive EMI — sparks, switching edges, or ESD-like pulses that mimic PD peaks.
  • Synchronous switching noise — bursts aligned to a known switching phase that pollute certain phase sectors.
  • External discharges — real discharges not originating in the DUT (fixture, connectors, nearby conductors).

Practical gating uses three families of evidence: phase-window rules (keep only physically plausible phase regions or mark suspicious sectors), band/feature rules (pulse width or energy signatures that separate PD-like pulses from interference), and cross-channel correlation (events appearing consistently across multiple sensors are more likely real). Gating should output a reason code for every rejected event so results remain reviewable.

Over-gating is a frequent failure mode: removing too aggressively can erase true PD and produce a deceptively “clean” PRPD. A safe practice is to keep summaries of rejected events (counts per reason, representative waveforms, phase distribution) so gating decisions can be validated.

Gating decision tree: phase, spectral or feature tests, and cross-channel correlation Decision tree showing input events evaluated by three criteria branches: phase consistency, band or feature signatures, and multi-channel correlation, leading to keep or reject decisions with reason codes and logging. Gating Decision Tree Reject EMI without hiding PD: explainable rules + reason codes Input events Detected pulses feature + phase + TOA Gate tests Explainable checks Phase window Plausible sectors phase consistency Band / features Pulse signature width · energy Correlation Multi-sensor TOA agreement Decision KEEP REJECT Evidence log Reason codes PHASE_FAIL FEATURE_FAIL Keep summaries of rejected events (counts, phase distribution, representative slices) to detect over-gating and preserve auditability.
Figure F6 — Gating rejects EMI using phase, feature, and correlation checks, while logging reason codes to keep decisions reviewable.

H2-7 · Calibration: charge injection, system transfer, and pC traceability

Apparent charge (pC) in PD testing is established through a transfer calibration: a controlled calibration pulse is injected at a defined point, the measurement chain response is captured, and a conversion factor is derived for the active range, bandwidth, and sensing path. The goal is not to claim a “true” microscopic charge distribution, but to keep results comparable and traceable under known conditions.

IEC 60270-style calibration loop (what must be controlled)
  • Injection element: the injection capacitor value/ID and its tolerance define the reference step.
  • Pulse conditions: amplitude, repetition rate, and edge behavior must be stable and documented.
  • Injection point: the point must follow the same coupling path used in measurement; a wrong point calibrates the wrong transfer.
  • Chain state: range state + bandwidth contract + sensor/path selection define the conversion factor in practice.

A practical instrument stores a calibration table indexed by configuration. A single “pC coefficient” is rarely sufficient because range switching changes gain and headroom, bandwidth settings change pulse shaping, and sensor placement or coupling path changes the effective transfer. Recording the calibration ID alongside the run configuration keeps PRPD, trending, and pass/fail decisions auditable.

Drift and re-check (keep calibration valid over time)
  • Temperature and warm-up shift gain and baseline, especially for wideband front ends.
  • Range state changes can introduce different offsets and compression behavior near clipping.
  • Sensor placement (HFCT position, coupling layout, cable routing) changes the effective path transfer.
  • Operational practice: run a quick power-on consistency check and schedule periodic verification under a known setup.

A reliable workflow keeps a compact “calibration snapshot” for every test: injection settings, chain state, and the calibration table entry used. This prevents silent drift from turning into false alarms or missed PD activity.

Charge calibration loop: pulse source, injection point, measurement transfer and indexed calibration table Diagram showing calibration pulse source and injection capacitor feeding a defined injection point, passing through the measurement chain to compute a pC coefficient, then writing results into a calibration table indexed by range state, bandwidth setting and sensor/path selection. Charge Calibration Loop (pC Traceability) Inject → capture transfer → compute coefficient → write indexed calibration table Cal pulse source Amplitude + rate Stable edges Injection C Value + ID Tolerance Injection point Same coupling path Point matters Measurement chain AFE + ADC + features Range + BW state Sensor/path select Compute pC coefficient Transfer response → scale Capture response Derive scale Calibration table (indexed) Range state · Bandwidth · Sensor/path Range BW Sensor pC coef R1 BW-A IEC K1 R2 BW-A HFCT K2 R2 BW-B UHF K3 Calibration snapshot Stored with every run CalID + Range + BW + Sensor Pulse settings + cable config Traceability comes from controlled injection + correct injection point + indexed coefficients + recorded configuration.
Figure F7 — A transfer calibration derives pC coefficients for each range/bandwidth/sensor condition and stores them with a run snapshot.

H2-8 · Test fixture & safety interlocks (HV enable, discharge, and fail-safe)

PD test fixtures must treat high voltage as a gated energy system. HV is enabled only when all interlock conditions are satisfied, and any fault forces the system into a de-energized state. A safe fixture also proves “safe-to-touch” through controlled discharge and residual voltage monitoring—because power-off does not guarantee zero stored energy.

Interlock chain (HV enable gate)
  • Access protection: door/lid closed, guard engaged, emergency-stop loop healthy.
  • Ground confirmation: protective earth continuity verified before enabling HV.
  • Connection checks: cable/connector presence and fixture state confirm correct routing.
  • Discharge readiness: bleed path and monitoring are operational before HV is allowed.

Discharge is part of the control loop: after HV is turned off or interlocks open, the fixture enters a controlled discharge state. A dedicated bleed path reduces stored energy, while a residual voltage monitor provides a positive indication of safe state. Clear status indicators (HV on / discharging / safe) prevent the dangerous assumption that “off means safe.”

Fail-safe and auditability (must-have behaviors)
  • Fail-safe de-energize: any interlock open forces HV inhibit and contactor open.
  • Controller fault behavior: loss of control power defaults to safe output states.
  • Event logging: every HV enable/disable is recorded with a reason code for review and maintenance.
Safety interlock chain with HV contactor, discharge path, and residual voltage monitoring Diagram showing a series interlock chain (door, e-stop, ground, cable, discharge ready) feeding a safety controller that enables an HV contactor to the DUT. A parallel bleed discharge path and residual voltage monitor provide safe-state indication, with event logging of HV enable/disable reason codes. Fixture Safety: Interlock → HV Enable → Discharge → Safe HV is gated by interlocks and forced off on fault; discharge and residual monitoring prove safe state Interlock chain (series) Door/Lid E-Stop Ground OK Cable OK Bleed OK Safety controller HV enable gate Fail-safe outputs HV contactor Open on fault DUT HV node Discharge and residual voltage proof Bleed path Controlled discharge Residual V monitor Safe-to-touch state Status HV ON · DISCHARGING · SAFE Event log Reason code Safety is a verifiable system behavior: gated HV enable, forced de-energize on fault, controlled discharge, and logged reasons.
Figure F8 — Interlocks gate HV enable; faults force de-energize. Discharge and residual monitoring provide a positive safe-state indication.

H2-9 · Measurement workflow: PDIV/PDEV, scanning, and reporting (PRPD + trends)

A repeatable PD workflow treats ramping, acquisition, and gating as controlled conditions, not operator preferences. PDIV/PDEV decisions become comparable only when the ramp rate, stabilization time, acquisition window, phase reference, and gating rule set are fixed and recorded with every run.

End-to-end workflow (inputs → outputs)
  1. Precheck gate: interlocks OK, discharge ready, calibration valid for active range/BW/sensor, and phase reference locked.
  2. Ramp: apply a fixed ramp rate and record it; comparisons across different ramp rates are not equivalent.
  3. Soak: wait for baseline and trigger stability before measuring; enter acquisition only after stable conditions are reached.
  4. Acquire: use a defined window length with a minimum event-count requirement and full phase-bin coverage to form a stable PRPD.
  5. Evaluate: apply PDIV/PDEV rules using consecutive windows and fixed gating; avoid single-window decisions.
  6. Report: export PRPD + pC statistics + event rate + gating ratio + setup snapshot + interlock log summary for auditability.

The acquisition window should be designed for statistical sufficiency: it must contain enough events to reduce randomness, enough phase coverage to avoid biased PRPD patterns, and a stable gating configuration so that “cleaning” does not shift the PDIV/PDEV boundary. A practical report always includes the gating ratio (rejected/total) and top reason codes to detect over-gating.

Reporting fields that preserve repeatability
  • PRPD plots: phase bins + amplitude bins + density scale.
  • pC statistics: peak, distribution summary, and event-rate trends over the scan.
  • Gating evidence: gating ratio and top reject reason codes.
  • Setup snapshot: range/BW/sensor, trigger rule ID, phase reference status, calibration ID.
  • Safety evidence: HV enable/disable reason summary and discharge-safe confirmation state.
End-to-end PD test timeline: precheck, ramp, soak, acquire, evaluate and report Timeline diagram showing a repeatable PD measurement workflow: precheck gate, ramp with fixed rate, soak to stabilize, acquisition window with phase lock and gating enabled, evaluation for PDIV/PDEV with consecutive windows, and reporting outputs. End-to-End Test Timeline Precheck → Ramp → Soak → Acquire → Evaluate → Report Precheck Interlocks OK Cal valid Phase locked Ramp Rate fixed Logged Soak Baseline stable Trigger stable Acquire Window + bins Gating on Count ≥ N Evaluate PDIV/PDEV Consecutive Report PRPD Trends Snapshot phase ref locked gating on cal valid Repeatability depends on fixed ramp rate, stable phase reference, fixed gating rules, and a recorded setup snapshot.
Figure F9 — A reproducible PD timeline highlights the control points that make PDIV/PDEV comparable and auditable.

H2-10 · Design checklist (what makes the PD front-end “good enough”)

A “good enough” PD front end is defined by measurable acceptance criteria. The checklist below turns bandwidth, noise, overload recovery, phase synchronization, gating evidence, and safety into pass/fail statements that can be verified during validation and commissioning.

Acceptance criteria (use as a verification gate)
  • BW / sampling: analog bandwidth and sampling rate meet the shortest target pulse width requirement.
  • Noise floor: integrated noise over the target BW supports the minimum detectable pC goal.
  • Dynamic range: gain/attenuation ranges cover weak PD while preventing long clipping on strong events.
  • Overload recovery: recovery time after large events does not bias event-rate statistics.
  • Trigger robustness: false trigger rate is controlled and rule sets are versioned and repeatable.
  • Phase sync: phase jitter/drift does not blur PRPD patterns across acquisition windows.
  • Gating evidence: gating output includes ratio + reason codes to detect over-gating and preserve auditability.
  • Interlock proof: fail-safe de-energize, verified discharge, and logged HV reasons are enforced by design.
Acceptance checklist card for PD front end: bandwidth, noise, recovery, sync, gating evidence and interlocks Checklist card diagram with 8 check items: bandwidth and sampling, noise floor, dynamic range, overload recovery, trigger robustness, phase sync, gating evidence, and safety interlocks. Acceptance Checklist Card Pass = measurable + repeatable + auditable BW / sampling Meets shortest pulse width target Noise floor Supports minimum detectable pC Dynamic range Weak PD visible, strong events handled Recovery No bias to event-rate statistics Trigger False triggers controlled + versioned Phase sync PRPD not blurred by jitter/drift Gating evidence Ratio + reason codes recorded Interlocks Fail-safe + discharge proof + logs Use this card as a release gate: if any item cannot be verified, PD results will not be comparable across setups and time.
Figure F10 — A compact acceptance card turns PD front-end quality into measurable verification gates.

H2-11 · Debug guide: symptoms → likely causes → what to probe/log

Field debugging starts with evidence, not assumptions. When “it looks like PD,” the fastest path is to link each symptom to a short list of high-probability causes and to capture the minimum set of waveforms, logs, and configuration snapshots needed to prove or eliminate each cause.

Capture-first rule (minimum evidence set)
  • Raw waveform snippets: pre-trigger + event + recovery window (for clipping and recovery checks).
  • Trigger threshold log: threshold, baseline estimate, drift over time, and false-trigger counters.
  • Gating output: gating ratio (rejected/total) and top reject reason codes.
  • Phase reference status: lock state, dropouts, and phase stamping quality.
  • Setup snapshot: range/BW/sensor-path, calibration ID/table index, and safety interlock events.
Symptom A — PRPD turns into a “blurred cloud” (no stable pattern)
Likely causes (priority order)
  1. Phase reference not stable: zero-cross jitter, lock dropouts, or drift in phase stamping.
  2. Over-gating / inconsistent gating: rule set changes across windows or removes too many events.
  3. Trigger drift: threshold/baseline moves, phase bins become effectively randomized.
  4. Timebase/sync mismatch: sampling/trigger domain not aligned with phase time-stamping domain.
What to probe / log
  • Phase lock status and dropout count (per acquisition window).
  • Phase stamp quality indicator (bin jitter / drift metric, if available).
  • Gating ratio + top reject reason codes (watch for “too clean” outputs).
  • Trigger threshold/baseline trend over time (drift correlates with PRPD blur).
  • Same-window PRPD phase coverage (ensure full 0–360° bin coverage).
Example parts (orientation only)
  • Phase/zero-cross comparator: TI LMH7220, ADI LTC6752, ADI ADCMP6xx family.
  • Isolation for phase/trigger I/O: ADI ADuM family, Silicon Labs Si86xx family.
  • Clock/sync distribution (if used): TI LMK family, ADI AD95xx family.
Symptom B — Event rate “jumps” with environment (nearby switching, movement, cable changes)
Likely causes (priority order)
  1. External EMI ingress: repeated narrowband interferers or random impulsive noise.
  2. Ground loop changes: loop geometry shifts, especially for HFCT/IEC coupling paths.
  3. Sensor mounting inconsistency: HFCT position/orientation/clamp force changes transfer.
  4. Cable routing / shielding changes: proximity and loop area variations alter coupling.
What to probe / log
  • Multi-channel correlation: does the same event appear across IEC/HFCT/UHF/TEV channels?
  • TOA clustering: impulsive EMI often arrives in bursts or with repeating cadence.
  • Frequency-band tags or summary spectrum features (even coarse classification helps).
  • Sensor installation snapshot code (position/orientation/route) per scan segment.
  • Gating reason codes: check whether rejects are dominated by “channel mismatch” or “phase-outside-window”.
Example parts / modules (orientation only)
  • HFCT sensor (example families): Pearson current monitors; Magnelab HFCT families (fixture-dependent).
  • Input protection (example class): TVS clamp families matched to front-end headroom (design-specific).
  • EMI filtering modules (example class): common-mode chokes / feedthrough filters (implementation-dependent).
Symptom C — pC readings change across range or bandwidth settings (inconsistent scaling)
Likely causes (priority order)
  1. Wrong calibration table entry: range/BW/sensor index not bound to the active chain state.
  2. Gain state not tied to CalID: switching range but reusing old coefficients.
  3. Filter configuration changed: analog/digital filtering reshapes pulses and affects feature mapping.
  4. Clipping not flagged: saturated events are still used, biasing statistics and scaling checks.
What to probe / log
  • Setup snapshot: range state, BW setting, sensor/path, and trigger rule ID.
  • Calibration: CalID + table index + timestamp of last verification.
  • Event feature consistency across ranges (peak / area / width trends on identical injections).
  • Clip/saturation flag per event (and count per window).
  • Processing version ID: filtering/feature-extraction version tag stored with the data.
Example parts (orientation only)
  • Programmable attenuator / gain block examples: ADI HMC attenuator families; ADI ADL gain blocks.
  • Range switching hardware examples: relay matrices (Pickering / Omron families) or solid-state analog switches (design-specific).
  • High-speed ADC families (system-dependent): ADI AD92xx/AD96xx families; TI ADC12/ADC14 families.
Debug log template (copy to test records)
  • Phase: ref source · lock status · dropouts · phase stamp quality
  • Trigger: threshold · baseline · drift · false-trigger counters
  • Gating: ratio · top reason codes · rule set ID
  • Chain: range state · BW setting · sensor/path · CalID/table index
  • Waveforms: pre-trigger length · clip flag rate · recovery-time evidence
  • Safety: interlock events · HV enable/disable reasons · discharge-safe confirmation
Troubleshooting map for PD measurement: symptom, priority checks, and evidence to capture Three-column troubleshooting map: symptoms on the left, priority checks in the middle, and evidence to capture on the right. Rows include PRPD blur, environment-driven event jumps, and range/BW scaling inconsistency, with short labels for each action. Troubleshooting Map (Evidence-First) Symptom → Priority checks → Evidence to capture Symptom Priority checks Evidence to capture PRPD blurred cloud Pattern unstable across windows Phase ref stability Gating ratio too high Trigger drift Phase lock + dropouts Reason codes + ratio Threshold/baseline log Event rate jumps Sensitive to environment/cables External EMI Ground loop geometry Sensor mounting Multi-channel correlation TOA clustering + tags Install snapshot code pC scaling inconsistent Changes across range/BW Cal table index binding Filter config drift Clipping not flagged Setup snapshot (range/BW) CalID + table index Clip flag + waveform If evidence is missing, postpone conclusions: capture waveforms + logs first, then re-run with fixed ramp rate and fixed gating rules.
Figure F11 — A three-column map keeps PD troubleshooting evidence-driven and repeatable in the field.

Request a Quote

Accepted Formats

pdf, csv, xls, xlsx, zip

Attachment

Drag & drop files here or use the button below.

H2-12 · FAQs ×12

Quick answers focused on wideband PD sensing, phase-synchronous PRPD, noise gating, calibration traceability, repeatable workflow, and safety interlocks.

Why is apparent charge (pC) not the “real charge” of the defect?

Apparent charge is an equivalent quantity defined by how a calibration pulse produces the same measured response through the coupling and measurement chain. It depends on the transfer function of the setup (sensor path, bandwidth, range, wiring), so it is not a direct measurement of the physical charge distribution inside a void or tree. Treat pC as a comparable indicator only within controlled conditions.

How should PDIV and PDEV decision rules be defined in practice?

Use fixed conditions (ramp rate, soak time, window length, gating rules, and phase reference) and define a threshold based on event rate and amplitude statistics. PDIV is reached when the threshold is exceeded for several consecutive acquisition windows; PDEV is reached when it remains below the threshold for several consecutive windows during a controlled down-ramp. Single-window calls are usually unstable.

How large should bandwidth and sampling rate be to avoid missing short PD pulses?

Start from the shortest pulse features that must be preserved (rise time or minimum width) and choose analog bandwidth high enough to avoid over-smoothing. Then select sampling rate to provide multiple samples across the relevant edge/feature and to support robust triggering and timestamping. Excessive filtering can erase pulse shape; insufficient sampling can distort amplitude, timing, and PRPD density.

Why does PRPD become “blurred,” and what phase-sync mistakes cause it most often?

PRPD blurs when phase stamping is unstable: phase lock dropouts, zero-cross jitter, drifting time alignment between the ADC stream and the phase reference, or inconsistent windowing across cycles. Over-gating can also smear patterns by deleting clusters selectively. Log phase-lock status, dropouts, and phase-stamp quality per window; then verify that phase bins cover 0–360° consistently.

How can gating reject EMI without deleting real PD events?

Use gating as an auditable decision, not a hidden filter. Combine multiple cues: phase-window constraints, frequency-band tags, and multi-channel correlation (events seen across independent sensors are harder to dismiss). Always store gating ratio (rejected/total) and reject reason codes. If gating ratio spikes or patterns disappear, reduce gating strength and re-check evidence with raw waveform snippets.

When are HFCT/UHF/TEV paths preferred versus an IEC 60270 coupling path?

An IEC-style coupling path is strong for controlled lab setups where the coupling network and measurement impedance are well-defined. HFCT is useful when ground-loop currents are accessible and installation is practical. UHF/TEV approaches can be effective in noisy environments or where near-field sensing helps separate external interference. Selection is driven by usable frequency band, noise immunity, and installation constraints for the DUT and fixture.

Why does front-end overload recovery determine whether event-rate statistics are trustworthy?

PD pulses can be sparse, but occasional large impulses can saturate amplifiers or ADCs. If recovery takes too long, the system becomes “blind” and undercounts subsequent events, biasing event-rate trends and PDIV/PDEV thresholds. Recovery should be measured and logged (clip flags, time-to-baseline, false triggers during recovery). Protection and gain staging should prevent long-duration clipping under expected worst-case impulses.

Where should the calibration pulse be injected, and how is traceability preserved?

Inject at a point that exercises the same coupling and measurement chain used during the test, so the transfer function is captured. Record the injection capacitor, pulse amplitude, repetition rate, cable/fixture configuration, active range, and bandwidth. Store a calibration ID and table index bound to each range/BW/sensor path. Re-verify after temperature changes, range switching, or sensor repositioning to prevent coefficient mismatch.

In strong external EMI, what evidence capture is most effective?

Prioritize evidence that distinguishes PD from interference: multi-channel correlation (events appearing across independent sensor paths), TOA clustering (bursts or repetitive cadence), and frequency-band tags (narrowband versus wideband impulses). Capture short raw waveform snippets around representative events, plus gating ratio and reason codes per window. A clean-looking PRPD without evidence is not reliable in high-EMI environments.

How should setup snapshots and interlock logs be included in reports for repeatability?

Store a compact “setup snapshot” with every scan segment: sensor/path selection, range and bandwidth, trigger rule ID, phase-reference status, and calibration ID/table index. Add safety evidence: HV enable/disable reasons and discharge-safe confirmation states. Include gating ratio and top reject reason codes alongside PRPD and trend plots. This makes runs reproducible and enables audit trails when PD boundaries are challenged.

How do ramp rate and stabilization time affect PD consistency and thresholds?

Ramp rate and soak time are measurement conditions. A fast ramp can push the system through transient states (baseline drift, phase-lock settling, trigger instability), causing inconsistent event rates and PDIV/PDEV decisions. A slow ramp improves settling but can introduce thermal drift and longer exposure to external disturbances. Fix ramp and soak parameters, require stable preconditions before acquisition, and record them in the report snapshot.

Which safety interlocks must be hard-series, and which can be monitoring/alarms?

Hard-series interlocks should remove energy by default: E-stop, enclosure/door/cover switches, grounding confirmation where applicable, and discharge-ready conditions that prevent HV enable. These should fail safe (open circuit → HV disabled). Monitoring/alarms can include residual voltage measurement, cable connection status, temperature, and “discharge completed” timing checks. Log every interlock transition with HV enable/disable reasons for traceability and incident review.

icnavigator

ICNavigator helps engineers find verified ICs, alternatives, and fast quotes.

Explore
Categories
  • Power Management
  • MCU
  • Automotive ICs
  • Sensors
  • Interface & Transceivers
  • Analog Front-End
Get in Touch
  • Email: hello@icnavigator.com
  • WeChat/WhatsApp: +86-xxxx

© 2025 ICNavigator. All rights reserved.