Hipot passes and the unit still fails in the field two years later. That pattern is almost always partial discharge: a void inside the insulation that withstands the short hipot test but erodes under continuous AC stress. For isolation transformers — medical, EV, telecom, instrumentation — hipot alone is not sufficient evidence of insulation health.
1. What each test actually measures
| Test | What it finds | What it misses |
|---|---|---|
| Hipot (AC or DC withstand) | Gross insulation breakdown, creepage failure, assembly error | Voids that do not break down at test voltage; long-term erosion |
| Insulation resistance (IR) | Moisture, contamination, tracking paths | Localised defects in series with a healthy path |
| Partial discharge (PD) | Voids, sharp electrode points, insufficient impregnation | Thermal and mechanical degradation without electrical symptoms |
| Surge / impulse | Transient withstand, isolation barrier margin | Continuous-stress ageing |
2. Choosing the test voltage
Standards define the withstand level: IEC 62368-1 for IT and AV equipment, IEC 60601-1 for medical (which also requires two means of patient protection where applicable), IEC 61558 for transformers generally. The practical rule is that the routine production test voltage is higher than the working voltage but below the level at which the test itself damages good insulation. DC hipot is often preferred for high-capacitance windings because it avoids the charging current that trips AC testers — but DC testing does not exercise the same stress mechanism as AC, which is why PD testing matters more for AC-stressed insulation.
3. Partial discharge: what to look for
PD measurement reports the apparent charge in picocoulombs at a specified voltage. Two numbers matter:
- PDIV — the voltage at which discharge starts. A well-impregnated isolation transformer should hold PD below the detection threshold at 1.2–1.5× working voltage.
- Extinction voltage — where discharge stops as voltage falls. A large gap between inception and extinction indicates a persistent void.
Typical acceptance for medical and EV isolation transformers is under 10 pC at the specified test voltage; general industrial parts often accept 50 pC. A reading in the hundreds of picocoulombs is a process problem, not a design problem — usually insufficient vacuum impregnation or a sharp winding transition point.
4. Where the failures come from
- Voids in impregnation. The single most common cause; cured by vacuum impregnation rather than by a thicker insulation wall.
- Sharp corners at the winding transition. Field concentration at a conductor corner erodes the enamel locally; use a margin tape or a radiused former.
- Creepage paths inside the bobbin.
- Insufficient layer insulation under high dv/dt. A 20 kHz square wave with 50 ns edges stresses inter-layer insulation far more than 50 Hz does.
- Contamination. Solder flux residue or handling contamination creates a tracking path that hipot may not catch if it is still dry.
5. Test sequence that catches real defects
IR at working voltage → PD at 1.2× working voltage → hipot at the standard level → IR again. Doing PD before hipot matters: once a hipot test has broken down a marginal void, the PD signature changes and the root cause is lost. Record PDIV and extinction voltage per lot, not just pass or fail.
6. When to require PD in the purchase specification
Require it for any part with a reinforced isolation barrier, any medical or EV application, any transformer above roughly 1 kV working voltage, and any design where the winding is not vacuum impregnated. For a 24 V signal transformer, hipot plus IR is adequate.
Shaanxi SiRui Technology tests Magnetic Core series isolation assemblies with IR, PD and hipot in sequence, and reports PDIV per lot so that a process drift is visible before it becomes a field return.
Send your isolation specification to sales@sirui-tech.com.