The DC hipot test is the most consequential offline test you can run on a stator winding. It applies a high DC voltage — well above operating levels — to the groundwall insulation and holds it there. If the insulation fails during the test, you know before the machine goes back into service. If it passes, you have reasonable assurance the groundwall can handle what the power system will throw at it.
IEEE 95-2002 is the governing standard: Recommended Practice for Insulation Testing of AC Electric Machinery (2300 V and Above) With High Direct Voltage. It applies to stator windings of AC machines rated 2300 V and above, covering both acceptance testing of new equipment and routine maintenance testing of machines already in service.
This article covers what the test actually does, how to select the right voltage, the three test methods and when to use each, how to read the results, and the safety procedures that the standard treats as mandatory.
Table of Contents
What the Test Is and Is Not
The DC hipot is an overpotential test — the applied voltage exceeds the peak of the normal operating voltage. The fundamental principle is straightforward: if there is a major flaw in the groundwall insulation, a high enough voltage will cause breakdown at that flaw. If the winding survives the test voltage without failure, it has demonstrated a minimum level of electrical strength.
There are two distinct uses of the DC hipot:
Proof testing is pass/fail. Apply the test voltage, hold it for the specified duration, observe whether the winding survives. The test result is binary: pass or fail. This is used for acceptance testing of new or rewound windings, and for maintenance testing of in-service windings.
Controlled overvoltage testing (stepped voltage or ramped voltage) is diagnostic. The voltage is increased gradually while continuously monitoring the leakage current. The shape of the current versus voltage curve reveals information about the insulation condition. In some cases, a developing problem can be detected and the test stopped before insulation breakdown occurs.
The standard is clear that both serve legitimate purposes, and both involve genuine risk: insulation failure during the test is possible, particularly during controlled overvoltage testing. Anyone performing the hipot test on a winding that has not been cleaned, dried, and confirmed to meet the minimum IR and PI thresholds from IEEE 43 is creating unnecessary risk of damaging insulation that could have been recovered.
Before the Test: Prerequisites
IEEE 95 is explicit on this point: before any high direct-voltage test, the stator winding must be deemed suitable for high voltage testing according to IEEE 43. The IR₁ and PI must be at or above the minimum values specified in IEEE 43-2000.
This is not a formality. Running a hipot on a wet or contaminated winding may puncture insulation that would have recovered after cleaning and drying. A winding that fails a hipot due to moisture contamination requires repair or rewind — a winding that failed the IR/PI test, was dried, and then passed the hipot may have needed nothing more than a cleaning.
Run IR/PI first. Always.
Additional pre-test checks required by the standard:
Temperature. The test should be made at winding temperatures at or below 40°C unless otherwise agreed with the manufacturer. Higher temperatures increase insulation conductivity and reduce apparent breakdown strength, which can make results inconsistent across tests conducted at different temperatures.
Humidity. Surface conduction current over endwindings increases significantly at high humidity, particularly where conductive contamination is present. The standard recommends keeping the winding temperature slightly above ambient (approximately 5°C above) to prevent condensation. Tests should ideally be made under humidity conditions similar to those the machine will operate in.
Surface contamination. Contaminants combined with moisture increase surface conduction current and can cause misleading results. Clean the winding before testing if contamination is suspected — particularly if oily or flammable contamination could present a fire hazard during flashover. The standard recommends testing both before cleaning (to detect incipient faults) and after cleaning and drying (for final fitness-for-service confirmation), where time and access permit.
External equipment. Surge arresters and surge capacitors must be disconnected before any DC hipot test. Surge arresters have resistive elements and gaps. Surge capacitors have internal discharge resistors, typically around 10 MΩ. Both will draw current in parallel with the winding and invalidate the measurements. Cables, bus insulators, and other connected equipment should be isolated where practical. Record what is connected in every test to ensure future tests are comparable.
Winding connections. Each phase should ideally be isolated and tested separately, with the other two phases grounded. Testing phases separately checks the phase-to-phase insulation — testing all phases simultaneously only checks the insulation to ground. For machines where separating the phases is impractical, the standard permits testing all phases together, with the understanding that phase-to-phase insulation is not assessed.
Test Voltages
Acceptance Proof Test
For new equipment tested at the factory or after field installation, the standard defers to the applicable equipment test code: IEC 60034-1, ANSI C50.10, and NEMA MG 1 Part 3. The DC acceptance proof test level is:
V_DC(acceptance) = 1.7 × V_AC(acceptance)
Where V_AC(acceptance) is the power frequency (50 or 60 Hz) acceptance test voltage defined in the applicable standard. The factor of 1.7 comes from studies comparing direct-voltage and power-frequency breakdown strengths on machine insulation — specifically, the ratio of DC to AC (rms) breakdown voltage for the slot portions of armature coils falls between 2 and 3, but for complete winding tests including endwindings and connections, 1.7 is used as a conservative factor that accounts for the weaker portions of the insulation system.
The standard acknowledges that the relationship between DC and AC breakdown strength is not precisely fixed — it depends on insulation type, age, and condition. The 1.7 factor is retained in the standard as a working value with an established empirical basis, not because it represents an exact equivalence.
Maintenance Proof Test
For in-service windings, IEEE 95 Section 6.2 states that an alternating voltage ranging from 125 to 150 percent of the rated rms line-to-line voltage E has proven adequate for maintenance testing. The DC equivalent is:
V_DC(maintenance) = 1.7 × V_AC(maintenance)
In practice, many users adopt a DC maintenance hipot level of approximately 2E — where E is the rated rms line-to-line voltage in kilovolts. For a 4160 V motor, this gives approximately 8 kV DC. The rationale: this level approximates the highest overvoltage the stator is likely to see in service if a phase-to-ground fault occurs in the power system. A winding that survives this level in a controlled, offline test is unlikely to fail in service from a voltage surge of similar magnitude.
Examples:
| Rated Voltage (rms) | AC Maintenance Test (125% of E) | DC Maintenance Test (1.7 × AC) | Common Practice (~2E) |
|---|---|---|---|
| 2300 V | 2875 V | 4888 V DC | ~4600 V DC |
| 4160 V | 5200 V | 8840 V DC | ~8320 V DC |
| 6600 V | 8250 V | 14,025 V DC | ~13,200 V DC |
| 13,800 V | 17,250 V | 29,325 V DC | ~27,600 V DC |
The test voltage for maintenance testing under special conditions — advanced insulation age, known damage, or previous anomalous results — may need to be adjusted. IEEE 95 recommends consulting the original equipment manufacturer in these circumstances.
Polarity
IEEE 95 specifies that negative polarity DC is historically preferred for high direct-voltage tests. This is because of electroendosmosis — the migration of moisture within the insulation toward the negatively charged electrode under the influence of an applied electric field. Negative polarity is considered more searching: it drives moisture toward the high-voltage conductor where it has more potential to reveal weakness.
If test results will be compared against previous or future tests on the same machine, the same polarity must be used throughout. The test record must note which polarity was applied.
Test Method 1: Proof Test (Pass/Fail)
The proof test is the simplest method. Apply the DC test voltage, hold for 1 minute, observe whether the winding survives.
Procedure:
- Complete all pre-test prerequisites — IR/PI check, isolation of external equipment, temperature and humidity assessment.
- Apply the DC voltage gradually. Do not exceed the test set’s maximum current rating during ramping, and avoid inadvertently tripping the overvoltage or overcurrent relays (which can introduce surges).
- Start timing when the target test voltage is reached.
- Hold for 1 minute.
- Reduce the output voltage control to zero.
- Discharge the winding through the discharge resistor until voltage reaches zero.
- Ground the winding solidly.
Pass/Fail: If no evidence of distress or insulation failure is observed during the full duration of voltage application, the winding has passed the proof test.
Insulation failure is typically indicated by a sharp capacitive discharge at the failure location. The test set circuit breaker may trip as the insulation resistance instantaneously drops to zero, causing the supply to attempt to deliver current it cannot supply. Erratic or large abnormal changes in measured current are also indicators.
If failure occurs, do not assume the winding is discharged. Treat it as energised until it has been properly discharged and grounded per the procedures in Section 6.4 of the standard.
Test Method 2: Stepped Voltage Test (Controlled Overvoltage)
The stepped voltage test applies voltage in increments while monitoring the leakage current at each step. It provides diagnostic information beyond a simple pass/fail result.
Two variants exist:
Uniform-time step: Voltage is increased in equal steps at regular time intervals — not exceeding 3 percent of the final test level per step — held for 1 minute each. Current is read at the end of each interval and plotted against voltage. For older asphaltic windings, the standard recommends 3-minute intervals to allow absorption current to decay further.
Graded-time step: A more sophisticated method that adjusts the time at each voltage step according to the absorption characteristics of the specific insulation system. The first step (30 percent or less of the maximum test voltage) is held for 10 minutes while IR, PI, and absorption ratio are measured and calculated. Subsequent step durations are read from a lookup table (IEEE 95 Table A.1) based on the measured absorption ratio. This method produces more accurate separation of conduction current from absorption current, but requires more calculation and discipline during execution.
Reading the current versus voltage curve:
A winding in good condition produces a smooth, rising curve — conduction current stays negligible relative to the geometric capacitive and absorption current components up to the maximum test voltage.
Warning signs in the stepped voltage test:
- Any deviation from a smooth curve at a specific voltage increment should be viewed as a possible warning of approaching breakdown. The standard recommends confirming the deviation with one additional voltage step before stopping.
- An accelerating rate of current increase — the curve bending steeply upward — is the most common indicator of approaching breakdown, particularly in windings at ambient temperature in humid air.
- Current that rises with time during a constant voltage hold (not just with the voltage step) indicates possible imminent breakdown.
- A very abrupt drop in conduction current above the peak operating voltage is rare but, when it occurs, indicates approaching failure. No method exists to estimate the breakdown voltage in this case — the standard recommends stopping the test immediately after confirming the phenomenon.
The standard is direct on stopping criteria: if the slope of the current doubles between two successive data points during the graded-time test, stop the test. If an extrapolation of the current curve suggests breakdown voltage is below the maximum planned test voltage, stop the test.
Test Method 3: Ramped Voltage Test
The ramp test continuously increases the voltage at a constant rate — typically 1 kV per minute — while recording current versus voltage with an X-Y plotter or digital recorder. The result is a continuous I-V curve rather than a series of discrete points.
The ramp test has several advantages over stepped methods:
- The continuous voltage rise automatically linearises the geometric capacitive and absorption current components, making deviations in conduction current more visible.
- Eliminating the human variable from reading meters, timing steps, and recording values at each increment produces more accurate and repeatable results.
- The slow, continuous voltage increase (approximately 1 kV per minute) is less likely to cause unpredictable insulation damage than the step increments of the conventional method (which can approximate 1 kV per second at each step).
- The ability to observe the developing I-V curve continuously provides better warning of approaching failure.
A well-regulated, stable power supply is essential for the ramp test. Any variation in the rate of voltage rise creates nonlinear changes in the capacitive and absorption current components and reduces the diagnostic value of the curve.
Reading ramp test results:
New epoxy-mica insulation: Produces a smooth, nearly linear I-V curve. Geometric capacitive current creates a positive offset. Absorption current contributes a gradual linear increase. Conduction current is negligible. This is the baseline signature of sound modern insulation.
New asphalt-mica insulation: Also linear, but with a steeper slope — the absorption current contribution is larger for asphaltic systems.
Localised weak spot (crack or void): A sudden, nonlinear increase in conduction current at a relatively low test voltage, well before the expected maximum. The curve departs sharply from linearity at that voltage. This is the signature of a single bad coil or localised flaw.
General deterioration (aged asphalt-mica): A gradual, progressive increase in the slope of the conduction current component as voltage increases. Not a sharp departure — a steady worsening. The standard describes this as evidence of many minute discontinuities contributing collectively to rising conduction current.
Moisture absorption (wet winding): Conduction current rises exponentially with applied voltage. The curve bends upward aggressively. The test should be stopped before the maximum planned voltage to avoid puncturing insulation that may be recoverable after drying.
Contaminated endwindings: Nonlinear current response but with a surface character — high conduction current at moderate voltages driven by surface leakage paths rather than bulk breakdown. The signature is different from groundwall cracking.
Delaminated insulation: A nonlinear I-V response with a characteristic shape — the slope increases sharply over a range of voltages, then levels off again. This has been associated with delaminated polyester-mica insulation. Detection of internal delamination by ramp test is possible but requires careful interpretation and is less reliable than for surface or crack-related problems.
Incompletely cured repair: Abnormal current response at lower voltages that disappears after the repair has had time to fully cure. After adequate curing time, the I-V curve returns to linear behaviour.
Discharging and Grounding
This is not optional detail — it is a safety requirement that IEEE 95 treats with explicit emphasis.
Upon completing the test, reduce the output voltage control to zero. This does not discharge the winding. A large winding holds significant stored charge — both capacitive charge (which discharges relatively quickly) and absorbed polarisation charge (which takes much longer to dissipate).
Discharge sequence:
- Reduce output voltage control to zero.
- Allow the winding voltage to decay through the series resistance of the test equipment to approximately half its previous value.
- Connect the discharge stick (with a resistor sized at 1000 to 6000 Ω per kV of maximum test voltage) to discharge the remainder.
- Once voltage is at zero, connect a solid ground — no resistor — to the winding terminals. Keep this ground in place.
Minimum grounding time after proof tests: The greater of 2 hours or 4 times the test duration — so after a 1-minute proof test, the minimum is 2 hours. For larger machines or longer test durations, the ground must remain in place longer.
The standard warns explicitly: if the ground is removed before the absorbed charge has fully dissipated, a recovery voltage will build up. This can be dangerous to personnel who touch the winding and can damage the insulation if the winding is energised or subjected to a subsequent test with residual polarisation charge present.
Do not attempt to accelerate dissipation by applying AC voltage or a reverse-polarity DC voltage. Both techniques introduce excessive internal voltage gradients and can cause insulation damage.
If an AC hipot test is to follow the DC hipot, the standard recommends doubling the minimum grounding time before the AC test — residual DC charge superimposed on the peak AC voltage could exceed the insulation’s breakdown strength.
Advantages of DC Hipot vs AC Hipot
IEEE 95 Annex A addresses this comparison directly. The DC test has practical advantages that explain its widespread field use:
The test set is compact, lightweight, and can operate from a standard 15 A supply. A power frequency AC hipot requires a large transformer — minimum approximately 170 kVA for a 4160 V motor — that is neither portable nor practical in most field settings.
Fewer partial discharges occur during DC testing compared to AC. DC does not age the insulation the way AC does — if the winding passes the DC proof test, the insulation has not been deteriorated by the test itself.
If breakdown does occur, the energy discharged into the fault is limited to the energy stored in the winding capacitance. This is much less damaging than an AC failure, where the transformer can supply sustained fault current.
The diagnostic information available from the current versus voltage response — particularly in stepped and ramped test variants — has no direct equivalent in a simple AC proof test.
The limitation the standard does not minimise: In the slot portion of the winding, the DC stress distribution follows resistivity and is similar to the AC stress distribution. In the endwindings, the distributions differ significantly. Under DC, the maximum surface voltage on the endwinding is reached at the maximum distance from the stator iron — opposite to what occurs under AC. This can create unnecessary stress on endwinding insulation near the core exit that is not experienced during normal operation. Some consider this a source of unnecessary failures during DC testing of specific winding configurations.
The other fundamental limitation — that DC testing cannot detect internal voids in modern mica-based insulation unless they have cracked through to ground — is shared with IR/PI testing and is inherent to DC test methods generally. AC-based methods, including power factor and partial discharge testing, are more sensitive to bulk internal conditions.
Fault Location
If the winding fails during the test, the fault must be found before the machine can be returned to service. IEEE 95 Section 8 provides several methods:
Reduce the test voltage immediately upon failure to prevent repeated flashover. Do not touch anything until the winding is properly discharged and grounded.
With the winding fully discharged and grounded, the location can be sought by:
- Applying a low AC voltage and increasing it slowly until the fault flashes to ground — locate the arc by sound, smoke, or light.
- For a low-resistance ground fault, using a clip-on ammeter at series connections to trace the fault to a specific slot.
- Measuring resistance from both the line end and neutral end of the faulted phase — the ratio of the two measurements indicates how far into the phase the fault is located.
- Probing with a grounded metal rod on an insulating stick to locate external endwinding faults.
- Infrared imaging to find heating at the fault site.
If the fault cannot be located by these means, sectionalize the winding — divide it and test each portion separately until the faulted section is isolated to a specific coil.
Documentation
IEEE 95 Section 9 defines the suggested test record. Every hipot test record should include:
Machine identification (station, serial number, manufacturer, type of insulation, rating); date, time, and duration of test; test voltage applied; leakage current at end of test; test connection arrangement and any connected apparatus; winding temperature and time at that temperature; ambient temperature and humidity; gas type and pressure if applicable; time out of service; test equipment description.
Also record: reason for test; visual inspection findings; IR and PI values from the pre-test IR/PI measurements; pertinent machine history; any observations of distress, corona, or unusual current behaviour during the test; result and action taken; recommendations for maintenance or future testing.
The benchmark value of the DC hipot is largely in its trend. A first-time ramp test on a new winding establishes the baseline I-V curve. Every subsequent ramp test is compared against that baseline. Changes in the shape or slope of the curve — even if the winding still passes the proof test voltage — are early warnings of developing deterioration.
Quick Reference
| Parameter | Value / Source |
|---|---|
| Standard | IEEE 95-2002 |
| Scope | AC stator windings rated 2300 V and above |
| DC acceptance test voltage | 1.7 × AC acceptance test voltage |
| DC maintenance test voltage | 1.7 × AC maintenance test voltage (125–150% of E) |
| Common maintenance practice | ~2E (DC) |
| Preferred polarity | Negative DC |
| Proof test duration | 1 minute at target voltage |
| Voltage step size (uniform-time method) | ≤ 3% of maximum test voltage per step |
| First step duration (graded-time method) | 10 minutes |
| Ramp rate (ramp test) | Typically 1 kV/minute |
| Minimum grounding time after proof test | Greater of 2 hours or 4× test duration |
| Prerequisite before test | IR₁ and PI must meet IEEE 43 minimums |
| DC vs AC internal void detection | DC cannot detect internal voids in mica-based insulation unless cracked through to ground |
| Effect on insulation | DC proof test does not age the insulation (unlike AC) |