Key takeaways
Partial discharge is an acceptance measurement, not an incoming inspection. Thermal cycling, vibration and damp heat push voids, burrs and fretting wear from undetectable to detectable; measuring before the stress is applied only verifies the moment assembly finished.
The conclusion first: partial discharge can prove that a defect exists, but not where it is; X-ray can give a location but not a mechanism; cross-sectioning can give a mechanism, but the sample is gone afterwards. These are not three interchangeable options - they form a chain with a fixed order. What is more often wrong is the timing: partial discharge is not an incoming inspection item but a repeat measurement taken after environmental stress. Measuring first and stressing afterwards only verifies the state at the moment assembly finished.
1. Correct the sequence first: PD belongs after the stress
Insulation defects in a high-voltage harness come from three places: burrs and voids in the crimp zone, air gaps between terminal and insulation, and mechanical damage at the interface between shield and jacket. All three share a property - immediately after assembly they are usually compressed, clamped by the jacket, or simply not open yet.
Thermal cycling makes dissimilar materials move against each other at their own expansion rates, opening the gaps; vibration produces micro-motion between terminal and crimp, wearing through plating and generating metal debris; damp heat lets moisture into the interface and distorts the local field. Only after that sequence does a gap too small to measure grow into a defect able to initiate discharge.
The order is therefore: baseline measurement, then the environmental stress items, then the acceptance measurement. The baseline proves the part was sound before; the acceptance measurement answers whether it still is. Without the second, a defect created by the stress is invisible; without the first, nothing can be attributed.
How environmental stress items are sequenced is covered in the ordering pitfalls of climate loads; restraint and fixing points for harnesses under vibration are in the harness vibration test article and on the vibration capability page. Those cover how stress is applied; this article covers how to inspect afterwards.
| Stage | What is measured | Why it has to be at this point |
|---|---|---|
| --- | --- | --- |
| After assembly | Baseline PD reading and background noise | Without a baseline, post-stress readings cannot be attributed |
| During environmental stress | Online monitoring (optional), visual and sealing checks | Anomalies during stress are closer to real service conditions |
| After stress | Acceptance PD reading, inception and extinction voltage | The defect has been amplified, so the reading becomes meaningful |
| After an abnormal result | X-ray / CT, then cross-sectioning and micro-sectioning | Non-destructive location first, destructive work last |
The stage split follows the general environmental test flow; specific items and their order follow the relevant volume of GB/T 28046 and the customer technical agreement.
2. Reading a PD result: three values beat one number
When a partial discharge report arrives, do not start with the peak value. Three other values carry more information.
The difference between inception and extinction voltage. Inception is the voltage at which stable discharge begins; extinction is the voltage at which it disappears as the level falls. When the two are close, the discharge sustains itself once started; a wide gap means the defect is unstable within a voltage window. For the same sample, a wider gap after stress than before means the defect has grown.
Repeatability across three voltage ramps. If inception voltage falls on each successive ramp of the same sample, the first ramp is itself enlarging the defect. That is not measurement error but defect evolution, and a report with a single ramp loses the information entirely.
The margin between sample reading and background noise. Measure the background in the chamber with no sample connected; the sample reading must sit clearly above it, otherwise any "exceedance" may be ambient interference. This item is the one most often omitted, and the one that most often triggers an audit finding.
As for the unit pC: it is only a dimension of charge, not a criterion. What counts as a failure comes from the vehicle manufacturer's specification clause or the technical agreement - the laboratory does not set that number on the customer's behalf. The method framework for partial discharge measurement is IEC 60270 and its national counterpart GB/T 7354; electrical and structural requirements for high-voltage harnesses themselves are in GB/T 37133 and the German LV 215 series, with revisions and clauses as stated in the standard text. Environmental stress items map to the ISO 16750 series and ISO 16750-2 electrical loads.
3. X-ray, CT and cross-sectioning: one says something is wrong, one says where, one says why
After an abnormal PD reading, the most common mistake is to cut the sample open straight away. Once cut, nothing is left for retesting or for a disputed review.
The correct order starts with non-destructive imaging: X-ray for voids in the crimp zone, terminal alignment and broken shield strands; CT when a three-dimensional position is needed. The image provides coordinates, and sectioning follows those coordinates so that the cut plane is targeted.
| Method | Can establish | Cannot establish | When it is used |
|---|---|---|---|
| --- | --- | --- | --- |
| Partial discharge measurement | Whether discharge exists, how inception and extinction relate, trend under stress | Defect position, defect type | Once before and once after stress, as the acceptance line |
| X-ray / CT | Crimp voids, terminal alignment, broken shield strands, internal gap position | Whether discharge is occurring, the discharge path | After an abnormal result, for non-destructive location |
| Cross-sectioning and micro-sectioning | Burrs, contamination, burn marks, tracking paths, plating wear | Statistical distribution across the batch | After location, for attribution and mechanism |
This division of labour is engineering practice rather than a mandatory standard sequence; the specific items follow the customer technical agreement. Sectioning and micro-sectioning belong to the materials side, and the materials and process analysis page lists the corresponding scope.
One coupling is frequently overlooked: broken shield strands are not only an insulation issue but also an EMC issue. Once shield integrity is compromised, radiated emission at higher frequencies degrades as well, and such samples usually look abnormal on the emission side at the same time as the PD reading. After finding broken strands, the emission items deserve a repeat run rather than an insulation repair alone.
Crimp and terminal behaviour on the connector side is covered by failure mechanism in automotive connector reliability; the difference in high-voltage applications is that the voltage level turns the same defect into a discharge risk, and the connector test item list maps the item ownership.
4. Four things the report must state
First, background noise and detection limit. Without these two numbers, nobody can tell whether the sample or the chamber was discharging.
Second, the source of the criterion. State which specification, which clause, which revision. A report that says "per industry practice" has no criterion at all under review.
Third, pre-stress conditions. Which items the sample had run before the PD measurement, in what order, and whether it failed in between. When readings from the same sample differ between two laboratories, this column is usually the misalignment. Differences between damp heat profiles are covered in cyclic versus steady-state damp heat.
Fourth, defect location and attribution. "Partial discharge exceeded the limit" is an observation, not a conclusion. The conclusion reads along the lines of: sample number so-and-so has a void at one crimp zone, and the micro-section shows plating wear and burn marks - attributed either to the crimping process or to the stress sequence.
A few plain words
The value of a partial discharge result is not in how large the number is, but in whether it can be reviewed. When TMC LAB takes on this kind of commission, the stress items and their order, the measurement points in time and the source of the criterion go into the test outline first, and after an abnormal result the evidence is preserved in the order of imaging location followed by sectioning attribution, so the sample is not destroyed before attribution is settled. Criteria and limits follow the specification clause and the technical agreement; no values are preset at the outline stage.
The genuinely time-saving move is to send the harness drawing, crimping process parameters and the specification clause number with the samples - far cheaper than retesting afterwards. Scheduling is arranged per project; you can state your expected delivery date when submitting samples, and we will confirm a feasible plan against the current schedule and report progress at each milestone.
