Common Mistakes in On-Site PD Testing: 10 Errors That Produce False Results
1. Testing without calibrating the measurement circuit
The mistake: Energizing the test setup and reading PD values without injecting a known calibration pulse first.
Why it fails: PD measurement is a calibrated measurement of apparent charge (pC). Without calibration you do not know the scale factor of the circuit — you are reading a voltage, not a charge. Cable length, coupling capacitor value and cable capacitance all change the transfer function.
The fix: Inject a known charge (typically 100 pC or 1 nC) with a PD calibrator at the test object, and verify the reading at the instrument. Record the calibration factor for every test. Re-check calibration if the test circuit changes (different cable length, different terminations).
2. Confusing external noise with PD
The mistake: Reporting PD activity that is actually corona from a nearby bushing, thyristor switching noise from the VLF source, radio interference, or a loose ground connection.
Why it fails: PD levels of interest (a few pC) are far below ambient noise in a substation. Noise that is not separated is either reported as a defect (false positive) or masks a real defect (false negative).
The fix: Record the background noise level before energizing the test object. Use PD measurement with noise gating / time-window separation, and compare the phase-resolved pattern of the signal against known PD patterns. If a signal appears before the test voltage is applied, it is not PD from the cable.
3. Using DC to excite PD in extruded cables
The mistake: Running a DC withstand test and expecting meaningful PD data on XLPE or EPR cable.
Why it fails: PD activity in extruded insulation is driven by the AC field distribution. Under DC, the field distributes by resistivity rather than permittivity, so the defect is not excited the way it is in service — and space charge from DC testing can itself alter later PD behavior.
The fix: Use VLF (0.1 Hz) as the PD excitation source for extruded cable — why VLF testing has replaced DC hipot. VLF is slow enough to be practical for field testing, but AC enough to reproduce service-like stress. For paper-insulated (PILC) cable the situation differs — but VLF is still the practical field choice.
4. Testing at the wrong voltage level
The mistake: Testing at a voltage that is too low to reach PD inception, or too high, creating PD that would not exist at operating stress.
Why it fails: PD has an inception voltage (PDIV) and an extinction voltage (PDEV). Below inception, a defective joint can look clean. Above the intended test level, you generate PD at the terminations that is an artefact of the test, not the asset.
The fix: Follow the voltage sequence defined by the applicable standard (typically rise to the specified test level, hold, then reduce to the PD measurement level). Measure at the specified level, not at whatever level the set happens to reach.
5. Taking a single reading with no trend
The mistake: Recording one PD value and declaring the cable good or bad.
Why it fails: PD is statistical. A single reading cannot separate a stable, harmless discharge from one that is actively deteriorating. The trend across tests — and the change in PD magnitude with voltage (the PD–voltage curve) — carries far more information than any single number.
The fix: Record the full PD–voltage characteristic (magnitude at each voltage step) and compare against the previous test on the same cable. Rising PDIV/PDEV margin loss over time is the real warning sign.
6. Not locating the PD source
The mistake: Knowing there is PD but not knowing where it is.
Why it fails: In field practice, the vast majority of PD originates at joints and terminations, not in the middle of an intact cable. A pC figure without a location cannot be acted on: you cannot plan a repair, and you cannot tell whether the signal is from the asset or from the test terminations.
The fix: Use PD mapping — time-of-flight and reflectometry techniques that locate the discharge along the cable — and correlate the reported distance with joint positions from as-built records. Any PD attributed to a joint should be verified against the joint's position before scheduling work.
7. Ignoring the test terminations as a PD source
The mistake: Assuming all measured PD comes from the cable under test.
Why it fails: Test terminations, temporary connections and adapters are common PD sources at test voltage. A test termination that is not properly assembled or is contaminated will generate PD that has nothing to do with the asset.
The fix: Keep test terminations clean, dry and correctly assembled. Test with a known-good termination where practical, and compare PD behaviour between the cable and the termination by testing at reduced voltage first.
8. Poor grounding and connection practice
The mistake: Long ground leads, daisy-chained grounds, or a ground connection with high impedance.
Why it fails: PD currents are small and fast. A high-impedance ground path distorts the measurement, adds noise, and can create apparent PD that is purely a circuit artefact.
The fix: Keep the PD measurement ground short and direct. Bond the cable sheath/system ground to a single low-impedance point. Verify the ground connection before every test.
9. Testing without a baseline from commissioning
The mistake: Running the first PD test on an asset that has decades of service and no historical PD data.
Why it fails: Without a commissioning PD baseline, you cannot tell whether a reading of, say, 50 pC represents long-term degradation or has been there since installation. Maintenance decisions then rest on thresholds alone.
The fix: Always record PD at commissioning — see PD testing for commissioning: what to measure before your new cable goes live — and store it with the TDR baseline trace and test certificates. Every later test is then a comparison, not a guess.
10. Treating PD testing as a one-off event
The mistake: Testing once, filing the report, and never revisiting the circuit.
Why it fails: PD is a precursor phenomenon. Its value comes from repeated measurement over the asset life: what changed, how fast, and which circuits are deteriorating.
The fix: Put significant circuits on a PD survey cycle (age, load, criticality based) and keep an online monitoring option for critical feeders. A single test finds today's defects; a programme finds tomorrow's failures.
Quick pre-test checklist
FAQ
Can PD testing produce false negatives? Yes — usually for three reasons: testing below PD inception voltage, not separating a real PD signal from background noise, or testing with DC instead of VLF on extruded cable. Any of these can make a defective joint look clean.
What PD level is acceptable in a new cable? For a newly commissioned extruded cable, the expectation is essentially no sustained PD activity above a few pC at the specified test voltage. Follow the acceptance criteria in the applicable standard and your utility's specification — but always record the actual value as a baseline, regardless of pass/fail.
Do I need PD mapping, or is a pC reading enough? If the reading is below the acceptance threshold, a value is usually sufficient. If PD is present, mapping is essential: without a location you cannot determine whether the source is a joint, a termination or test equipment, and you cannot plan work.
How often should on-site PD testing be repeated? For critical or elderly circuits, an annual or biennial survey is common practice (see how utilities run PD surveys across a network); circuits flagged with PD activity should be retested sooner to establish the trend. The interval should be driven by how fast PD is changing, not by a fixed calendar alone.
Related reading
- Partial Discharge Testing of Power Cables — what PD is, and why it predicts failure years in advance
- PD Testing for Commissioning — what to measure before a new cable goes live
- PD Testing Equipment for Commissioning — the four functional blocks every test kit needs
- How to Interpret Partial Discharge Test Results — magnitude, PDIV/PDEV and pass/fail criteria