Partial discharge is the acceptance criterion that separates a cast resin transformer built to last thirty years from one that fails in year six, and it is the one number in the test report that a buyer can check without any test equipment. For dry-type transformers to IEC 60076-11, the commonly specified limit is not more than 10 pC measured at 1.1 times the rated system voltage divided by √3. A unit that measures 3 pC and a unit that measures 90 pC can both pass a routine ratio and insulation resistance test — only the PD result tells them apart.
This guide explains what partial discharge is in a cast resin winding, where the limits come from, how the test is actually run, what the reading means at different values, and what to do when a unit misses the limit. The physical construction that determines PD performance is covered in the guide to dry-type power transformers, and specifically in the article on cast resin construction.
Partial discharge is a localised electrical breakdown that does not bridge the full insulation between two electrodes. It happens inside a gas-filled void, along an interface between materials, or at a sharp electrode edge, and it repeats at every half cycle. Each event erodes a few molecules of insulation and deposits decomposition products. The damage is cumulative, and the mechanism accelerates: a void that discharges grows, and a growing void discharges harder.
A cast resin winding is a composite — copper conductors, a foil or wire winding, interlayer insulation, cast epoxy with silica filler, and the interfaces between all of them. Every interface is a potential cavity. The epoxy cures with shrinkage that can pull it away from the conductor. Thermal cycling from load variation opens micro-gaps. A sharp conductor corner concentrates field stress. None of these defects shows up in a ratio test, a winding resistance measurement, or an insulation resistance reading, because the insulation is still intact — it is only partially bridged.
Oil-immersed transformers have a different exposure. The oil impregnates and fills voids, and the PD mechanism is dominated by gas bubbles generated by overheating or by moisture. That is why the limits differ: dry-type limits are set in the low tens of picocoulombs while oil-immersed acceptance limits are typically two orders of magnitude higher.
IEC 60076-11 is the reference standard for dry-type power transformers and it defines a partial discharge level that the winding must not exceed during the routine dielectric tests. The figure is 10 pC, and it is quoted at the test voltage of 1.1 × Um/√3, where Um is the highest voltage for equipment.
| System voltage Um | PD test voltage (1.1 Um/√3) | Standard limit | Tightened specification |
|---|---|---|---|
| 7.2 kV | 4.6 kV | 10 pC | 5 pC |
| 12 kV | 7.6 kV | 10 pC | 5 pC |
| 24 kV | 15.2 kV | 10 pC | 5 pC |
| 36 kV | 22.9 kV | 10 pC | 5 pC |
| 40.5 kV | 25.7 kV | 10 pC | 5 pC |
Two practical notes. First, the 10 pC figure is a ceiling, not a target — a well-made cast resin winding typically measures in the 2 to 6 pC range, and a supplier reporting “less than 10 pC” on every unit is reporting a pass, not a quality level. Second, the exact test sequence and any preconditioning voltage should be read from the edition of the standard named in the purchase order, because editions differ and because the buyer’s specification often adds requirements above the baseline. The surrounding test programme is described in the factory acceptance test guide.
The measurement itself uses the IEC 60270 method: a coupling capacitor and a measuring impedance in the test circuit, with the discharge magnitude calibrated by injecting a known charge pulse before the test. The calibration step is what makes the picocoulomb figure meaningful; without it, the reading is just a number on a dial.
The test is normally carried out alongside the induced overvoltage test and the applied voltage test, since the voltage source and the shielded room are already set up. Doing PD as a standalone test on a finished unit means shipping it back to a test bay if it fails, which is why it belongs in the routine sequence rather than the site acceptance scope.
| Measured level | Interpretation | Action |
|---|---|---|
| Under 5 pC | Clean insulation system, well impregnated, void-free interfaces | Accept; typical of good cast resin work |
| 5 to 10 pC | Minor voids or interface activity present but within standard | Accept for general distribution duty; query if the specification is 5 pC |
| 10 to 50 pC | Beyond the IEC limit; a real defect exists somewhere | Reject, or require investigation and retest after treatment |
| 50 to 200 pC | Significant cavity or delamination; will propagate in service | Reject; the winding is not repairable without rework |
| Above 200 pC | Severe defect, possibly a crack or a contaminated interface | Reject; investigate the casting process, not just the unit |
A number alone is not the whole story. A unit that measures 8 pC but whose inception voltage is barely above the operating voltage has very little margin, because any voltage rise in service will push it into sustained discharge. A unit that measures 8 pC with inception at twice operating voltage is comfortable. If the specification allows it, ask for inception and extinction voltages alongside the magnitude.
PD in a cast resin transformer almost always traces to one of six causes, and the pattern in the test data often identifies which.
The remedy is always in the process, not the test. A supplier whose units routinely measure 20 to 40 pC has a vacuum casting or cure problem, and tightening the acceptance limit without changing the process simply moves the rejects to a later stage. This is why PD is described as a process indicator as much as a product test — the same logic that makes loss measurement an indicator of design quality, covered in the article on transformer efficiency classes and losses.
PD does not replace the other tests, and the other tests do not replace PD. The useful picture comes from reading them together.
| Test | What it catches | What it misses |
|---|---|---|
| Partial discharge (IEC 60270) | Voids, delamination, field enhancement | Winding geometry errors, connection mistakes |
| Insulation resistance / PI | Moisture, gross contamination, tracking paths | Localised defects in sound insulation |
| Applied voltage withstand | Gross insulation weakness that fails outright | Defects that survive the one-minute test |
| Induced overvoltage | Turn-to-turn and phase-to-phase weakness | Defects not stressed by the test voltage profile |
| Ratio and polarity | Winding errors, tap connection mistakes | Anything related to insulation quality |
| No-load and load loss | Core and winding design or assembly problems | Insulation defects entirely |
The full specification of what a routine test report should contain, and how the values relate to the design data sheet, is set out in the guide to the dry-type transformer specification.
Factory PD says nothing about what happens after transport, and transport is where cast resin units take damage. A crack initiated by a lifting sling or a poorly blocked shipment will not appear in a visual inspection but will show as PD once energised. For critical installations — data centres, hospitals, tunnel substations — a site PD measurement before handover closes that gap.
Onsite PD measurement is harder than factory measurement because the environment is electrically noisy. Online methods use high-frequency current transformers on the earth connection, ultra-high-frequency sensors, or acoustic detection, and each trades sensitivity for practicality. A common approach is a baseline measurement at commissioning, repeated annually, so that a trend is available before a failure. Absolute values are less useful than the trend, because site conditions vary.
The insulation system that determines how much PD margin a unit has in the first place is covered in the guide to dry-type transformer insulation, including the temperature class and material choices that interact with the casting process.
Treat the partial discharge figure as a process signature rather than a pass mark. Ask for the measured value, not just a compliance statement; ask for the test voltage and the standard edition; and treat a consistent 8 to 9 pC reading across a whole order as a warning about casting control, not as evidence of quality. On a specification where the winding has to survive decades of thermal cycling, a few picocoulombs of margin is worth more than most of the other line items on the quotation.
IEC 60076-11 sets 10 pC at 1.1 times Um/√3 as the limit for cast resin dry-type transformers. Well-controlled production typically measures 2 to 6 pC; many specifications tighten the acceptance level to 5 pC.
It caps the charge transferred in each individual discharge event at 10 picocoulombs — an extremely small quantity. Anything above it indicates a cavity, delamination or field concentration large enough to erode the insulation over time.
Not immediately, but it is progressive. Discharge erodes insulation and generates decomposition products that increase local stress, so a unit operating above its inception voltage degrades faster than the design life assumes.
Oil fills and impregnates voids and recovers from local breakdown, so the discharge mechanism and its consequences differ. Dry-type insulation has no self-healing medium, which is why its limits are set two orders of magnitude lower.
Sometimes. Surface contamination or moisture can be removed by cleaning and drying. An internal casting void or delamination cannot be repaired without rewinding the coil, which is rarely economic.
It is a routine test on every dry-type unit for most specifications, not a type test. Type tests establish the design; the routine PD measurement confirms each individual winding.
