Insulation resistance testing is one of the most practical condition assessment tools in electrical maintenance. Apply DC voltage to a winding, measure the leakage current, divide one by the other — you get a number in megohms that tells you whether the insulation is clean and dry, wet and contaminated, or failing. IEEE 43-2013 is the standard that defines exactly how to do it and how to read the result.
This article covers the full procedure: the physics behind the measurement, how to select the right test voltage, how to correct for temperature, minimum acceptable values, and — critically — where the test’s limits are. Knowing what it cannot find is just as important as knowing what it can.
Table of Contents
What the Test Actually Measures
When you apply DC voltage across a motor winding to ground, four distinct currents flow through and across the insulation simultaneously. The megohmmeter measures their combined total. Understanding each current separately is what lets you interpret results properly.
Geometric capacitive current (I_C) charges the winding capacitance. It is large at the start and decays exponentially within the first few seconds. It has no diagnostic value. This is why you wait a full minute before reading — by then, this current is negligible.
Absorption current (I_A) arises from molecular polarization at the interfaces between the different materials in the insulation system — mica, glass, epoxy, polyester. It decays over many minutes following a power-law relationship (I_A = K × t^-n, where n is a characteristic of the insulation system). In clean, dry insulation, this current dominates the measurement between 1 and 10 minutes. In contaminated or moisture-saturated insulation, it is swamped by surface leakage current and the diagnostic value of the time-based tests disappears.
Conduction current (I_G) passes through the bulk of the insulation and is constant with time. In modern epoxy-mica windings it is essentially zero — the material is a near-perfect insulator at DC voltages. In older asphaltic systems, or in any insulation that has absorbed moisture deep into its bulk, this current is measurable and degrades the IR reading directly.
Surface leakage current (I_L) flows across the insulation surface along the end turns, driven by semi-conducting contamination — oil, carbon dust, moisture, salt deposits. It is constant with time. When it dominates, the total current barely changes over 10 minutes, the Polarization Index stays close to 1.0, and the winding needs cleaning before any further test result is meaningful.
The megohmmeter cannot separate these currents. What the test procedure and the Polarization Index do is use the time behaviour of the total current to infer which components are dominant — and from that, what condition the insulation is in.
Scope of IEEE 43-2013
The standard applies to armature and field windings of rotating machines rated 750 W (approximately 1 HP) or greater. This includes:
- AC induction motors and synchronous motors
- Synchronous generators
- DC machines — both armature and field windings
- Synchronous condensers
It does not apply to fractional-horsepower machines. Squirrel-cage induction rotor windings are excluded because they have no insulation to test. The PI test is also not applicable to DC machine armatures with exposed copper commutators and to non-insulated field windings — reasons covered in detail below.
Test Voltages
Selecting the right DC test voltage is the first critical step. Too low and you will not stress the insulation enough to reveal weaknesses. Too high and you risk damaging a borderline machine — particularly one that is wet or contaminated.
IEEE 43-2013 Table 1 gives the following guidelines:
| Winding Rated Voltage | Recommended DC Test Voltage |
|---|---|
| Below 1000 V | 500 V DC |
| 1000 V to 2500 V | 500 to 1000 V DC |
| 2501 V to 5000 V | 1000 to 2500 V DC |
| 5001 V to 12,000 V | 2500 to 5000 V DC |
| Above 12,000 V | 5000 to 10,000 V DC |
Rated voltage in this table is line-to-line for three-phase AC machines, line-to-ground for single-phase machines, and rated direct voltage for DC machines and field windings.
One specific note from the standard: machines with stress control coatings on the endwinding can have their resistance and PI measurements distorted at lower test voltages, because the stress grading material influences the current profile. For these machines, test voltages of 5 kV and above produce more reliable results — the stress grading system’s contribution to the measured current diminishes at higher voltage levels.
The standard also specifies that tests should be conducted using negative polarity DC. This accommodates the electroendosmosis effect, which can produce different readings depending on polarity in older thermoplastic asphaltic windings that have absorbed moisture.
Pre-Test Requirements
The winding must be at zero residual charge before testing. Residual charge from a previous test or from the machine having been energised distorts the measurement — this is the memory effect described in the standard. The practical check is to measure the discharge current at the start of the test. If it is still above background level, the winding has not discharged sufficiently.
Record the following at every test:
- Ambient temperature
- Relative humidity and dew point
- Winding temperature
- Time out of service
- Test voltage applied
- Connection arrangement (all phases together or individual phases)
These records are not optional extras. They are what allow you to compare this result against the one from a year ago. Without them, you have an isolated number with no context.
For directly water-cooled windings, the cooling water must be drained and the internal circuit thoroughly dried before testing. Some manufacturers provide testing arrangements that do not require draining — but only where cooling water conductivity is below 0.25 µS/cm. Confirm this in the machine documentation.
Test Connections
IEEE 43 recommends testing each phase individually where feasible. When one phase is under test, the other two phases should be grounded to the same ground as the stator core or rotor body. Individual phase testing allows direct comparison between phases and can reveal asymmetric problems.
When all three phases are tested together — which is common when individual neutral access is not available — the test measures phase-to-ground insulation only. No phase-to-phase information is obtained.
Disconnect and ground all external equipment before testing: surge capacitors, surge arresters, cables, instrument transformers, anything connected to the motor terminals. A surge capacitor left connected to the terminals will give a PI reading close to 1.0 regardless of winding condition, because the capacitor’s internal discharge resistor (typically around 10 MΩ) will dominate the measurement and mask the absorption current behaviour. If something cannot be disconnected, note it in the record for future comparisons.
The One-Minute IR Test
Apply the DC test voltage and read the insulation resistance at 60 seconds. That reading is IR₁.
The insulation resistance at time t is calculated as:
IR_t = E(t) / I(t)
Where E is the applied voltage in volts and I is the total current in microamperes. The result is in megohms.
After taking the reading, remove the high-voltage lead from the terminal and immediately ground the winding through a suitable resistor sized to limit instantaneous discharge current. The minimum discharge time is four times the test duration — at least 4 minutes after a 1-minute test. Do not lift the ground until discharge current is essentially zero and return voltage is below approximately 20 V. IEEE 43 is explicit: the test is not complete until the winding is discharged. Residual charge on a large winding can persist for hours and constitutes a genuine shock hazard.
The Polarization Index
The PI is measured in the same test session as IR₁. Apply voltage for 10 minutes. Read at 1 minute and at 10 minutes. The PI is:
PI = IR₁₀ ÷ IR₁
The PI works largely independently of temperature. Both readings are taken at essentially the same winding temperature, so the temperature correction factor that applies to each is nearly identical — when you divide one by the other, it cancels. This is the PI’s main advantage over a raw IR₁ reading for comparing tests across different ambient conditions.
What the PI reveals is whether the absorption current is declining over time — which it does in clean, dry insulation as the molecular polarisation settles — or whether the total current is approximately flat because surface leakage or conduction current is dominating.
IEEE 43-2013 Table 3 gives the minimum recommended PI values based on insulation thermal class:
| Insulation Thermal Class | Minimum Recommended PI |
|---|---|
| Class 105 (A) | 1.5 |
| Class 130 (B) and above | 2.0 |
Most industrial motors use Class F (155°C) or Class B (130°C) insulation, so the minimum PI of 2.0 applies in the majority of cases.
When the PI Does Not Apply
The PI has clearly defined limitations. The standard addresses all of them directly.
When IR₁ exceeds 5000 MΩ. At this resistance level, the total current is in the sub-microampere range. At that sensitivity, small fluctuations in supply voltage, ambient humidity, test lead surface condition, and nearby metallic objects significantly affect the measured current between the 1-minute and 10-minute readings. The resulting PI may or may not reflect actual insulation condition. IEEE 43 states explicitly that when IR₁ exceeds 5000 MΩ, the PI is not a reliable assessment tool and need not be calculated.
Non-insulated field windings. Large turbine generators and some salient pole machines use strip-on-edge field windings where individual conductors are not encapsulated in insulation — they are isolated from ground by insulating strips and barriers only. The exposed copper surface area is enormous relative to the insulated surfaces. Surface leakage current (I_L) dominates over absorption current (I_A), so the IR curve stays flat over time. The PI will always read close to 1.0 regardless of actual condition. The standard is explicit: PI testing is not applicable to non-insulated field windings.
DC machine armatures with exposed commutators. The commutator surface area creates the same problem — surface leakage dominates. PI is not applicable.
Continuous stress control coatings. Some Roebel-bar windings have the entire endwinding overhang treated with stress control material that makes electrical contact with the bare copper conductor ends. The surface leakage this creates can drive the PI close to 1.0 even when the insulation is in perfect condition. The standard flags this as a design characteristic, not a fault — but notes that incorrect application of this system can eventually lead to electrical tracking.
Small random-wound machines. The absorption current in small random-wound motors may decay to essentially zero in 2–3 minutes rather than the 10 minutes the standard test assumes. For these machines, the standard describes variant PI calculations using shorter time intervals — IR₁/IR₃₀ₛ or IR₅/IR₁ — as alternatives that save test time without losing diagnostic information. No standardised pass/fail criteria have been established for these variants yet.
Minimum Acceptable Insulation Resistance
This is where IEEE 43-2013 made a significant change from older guidance — and where a lot of field guides are still using outdated thresholds.
IEEE 43-2013 Table 4 defines three separate minimum IR₁ values:
| Minimum IR₁ at 40°C | Application |
|---|---|
| kV + 1 MΩ | Most windings made before about 1970, all field windings, and others not described below |
| 100 MΩ | Most AC windings built after about 1970 (form-wound coils) |
| 5 MΩ | Most machines with random-wound stator coils, form-wound coils rated below 1 kV, and DC armatures |
Where kV is the rated line-to-line voltage in kilovolts for three-phase AC machines, line-to-ground for single-phase, and rated direct voltage for DC machines and field windings.
The split between pre-1970 and post-1970 windings is the critical distinction. A modern 4160 V motor with epoxy-mica form-wound coils has a minimum IR₁ of 100 MΩ — not the 5.16 MΩ you would get from the kV + 1 formula. Applying the old formula to a modern winding sets the acceptance threshold approximately 20 times lower than it should be.
The reason is material behaviour. Modern epoxy-mica insulation has essentially infinite bulk resistivity when clean and dry. Healthy modern form-wound windings routinely read in the hundreds of MΩ or into the GΩ range. The standard’s 100 MΩ minimum reflects field experience that readings below this level, even on modern systems, indicate contamination or moisture that warrants investigation before energising.
Older asphaltic mica systems genuinely behave differently — they absorb moisture more readily, their bulk resistivity is lower, and their IR values trend lower even in acceptable condition. The kV + 1 formula was calibrated to that class of material.
All IR₁ comparisons against these minimums must use the value corrected to 40°C.
Temperature Correction
Insulation resistance varies exponentially with temperature. As temperature rises, thermal energy frees additional charge carriers in the insulation material and resistivity falls. A motor winding at 20°C will give a significantly higher IR reading than the same winding at 40°C — not because the insulation has changed, but because temperature changed.
IEEE 43-2013 sets 40°C as the standard reference temperature. The correction formula is:
R_C = K_T × R_T
Where R_C is the corrected resistance at 40°C, R_T is the measured resistance at temperature T°C, and K_T is the temperature correction factor read from the appropriate curve for the insulation type.
The standard defines two insulation families with separate correction equations and curves:
Thermoplastic systems — asphaltic-mica and shellac mica-folium, generally pre-1960s machines:
K_T = (0.5)^((40−T)/10)
Thermosetting systems — epoxy and polyester based systems, generally post-1960s. Two equations apply depending on whether the winding temperature is above or below 40°C.
IEEE 43-2013 Table 2 gives the computed K_T values for both families:
| Temperature (°C) | K_T Thermoplastic | K_T Thermosetting |
|---|---|---|
| 10 | 0.125 | 0.7 |
| 20 | 0.25 | 0.8 |
| 30 | 0.5 | 0.9 |
| 40 | 1.0 | 1.0 |
| 50 | 2.0 | 1.5 |
| 60 | 4.0 | 2.3 |
| 70 | 8.0 | 3.3 |
| 80 | 16.0 | 4.6 |
Two things stand out. First, thermoplastic insulation is far more temperature-sensitive than thermosetting. At 60°C, the thermoplastic correction factor is 4.0, meaning you multiply the measured reading by 4 to get the 40°C equivalent. The thermosetting correction at the same temperature is only 2.3. Applying the thermoplastic curve to a modern epoxy winding would significantly over-correct the result.
Second, correction breaks down below the dew point. When moisture is condensing on the insulation surface, the correction factors do not apply — they were established on clean, dry test specimens. The standard recommends that in this situation, the machine’s test history under similar conditions should be the primary reference, rather than a temperature-corrected absolute value.
Temperature correction is not required for the PI — the ratio cancels K_T. One exception: if the winding temperature is falling significantly during the test (a machine recently stopped from a high operating temperature), the IR may rise during the 10-minute PI test due to temperature drop rather than purely polarisation behaviour. The resulting PI can be artificially high. If this is suspected, repeat the test after the winding has cooled below 40°C.
Effect of Test Voltage Magnitude
For insulation in good condition and thoroughly dry, the IR will be approximately the same regardless of which test voltage within the recommended range is applied. This is expected behaviour.
A significant decrease in IR as applied voltage increases is a flag — it indicates imperfections or fractures in the insulation, often aggravated by moisture or contamination at the defect. The standard notes this effect becomes more pronounced at voltages well above rated voltage. This is the physical basis of the step voltage test in IEEE 95. It is also why the test voltage must be recorded in every test session — if the voltage changes between sessions, the IR values are not directly comparable even after temperature correction.
Interpreting Results
IEEE 43 describes two situations that require different approaches.
When historical data is available. Compare the current test against previous tests under similar conditions — same test voltage, similar winding temperature, similar humidity. Correct all readings to 40°C before comparing. The trend is the primary diagnostic signal. A sharp drop in IR₁ or PI from the previous reading is a concern regardless of whether the absolute reading still clears the minimum threshold. Conversely, the standard notes that a steady increase in IR₁ on older thermoplastic windings over time can itself indicate thermal decomposition of the bonding materials — because aged asphaltic insulation loses its hygroscopic character, the absorption current decreases, and a very high PI (above 8) on an old asphaltic winding may signal that the insulation is dry and brittle and at elevated risk of mechanical failure.
When no historical data exists. Use the minimums from Table 3 (PI) and Table 4 (IR₁ corrected to 40°C). Both must be above their respective minimums. The standard requires both — a machine with adequate IR but low PI, or adequate PI but IR below the minimum, does not meet the criteria for energisation or overvoltage testing.
Limitations of the IR Test
IEEE 43-2013 Section 11.3 lists the limitations of insulation resistance testing. These are not footnotes — they are fundamental to knowing when to use additional tests.
IR is not a measure of dielectric strength. A specific IR reading does not tell you at what voltage the insulation will break down. The test cannot predict whether a winding will survive a hipot test or how many years of operation remain.
The test cannot locate defects. One number covers the entire winding under test. A single deteriorated coil can pull down the phase IR without any indication of where the problem is or how localised it is.
The test cannot detect internal voids in modern epoxy-mica insulation. This is the most critical limitation for modern form-wound stator coils. Mica has virtually infinite DC resistivity. Even a single layer of mica tape is sufficient to block all direct current. A void or delamination within the groundwall insulation — from thermal deterioration, thermal cycling, or incomplete vacuum pressure impregnation — will not be visible in an IR or PI reading. As Annex B of IEEE 43 states, because mica essentially blocks all DC conduction, only a severe crack that provides a continuous conductive path from conductor to ground will register. Internal defect detection in modern windings requires AC-based testing: power factor testing, capacitance tip-up, or partial discharge testing.
The test cannot detect rotation-related problems. Loose coils, endwinding vibration and fatigue cracking, and coil loosening in the slot are not detectable when the machine is at standstill.
Large or slow-speed machines may not reach recommended minimum values. Machines with very large endwinding surface areas have inherently high surface leakage current paths. Their IR₁ may be below the standard minimums even in good condition. For these machines, historical trending of IR₁ under consistent conditions is the appropriate methodology.
What to Do When Results Are Below Minimum
Do not energise a winding that has failed the IR₁ or PI minimum. Investigate first.
Low IR₁ and low PI (near 1.0): Surface contamination or moisture is the likely cause. The insulation is not necessarily damaged. Dry the winding — forced warm air, space heaters inside the enclosure, or operating uncoupled and unloaded — then retest. The PI can monitor the drying progress: continue until the PI exceeds the minimum and stabilises over successive tests. The standard specifically notes that PI can be used to determine when drying is sufficient.
Low IR₁ but PI in acceptable range: The low reading may be driven by surface condition. Cleaning and drying should help. The acceptable PI suggests the bulk insulation is polarising normally.
Both low and not recovering after drying: The insulation has degraded beyond moisture effects — cracking, thermal deterioration, or deep contamination. Further investigation is needed before any repair decision: power factor testing, hipot, or visual inspection of accessible surfaces.
The standard explicitly recommends that a machine with both low PI and low IR₁ should not be subjected to further high-voltage testing. Conducting a hipot on a wet or contaminated winding risks puncturing insulation that would have been recoverable after drying.
Insulation Resistance Profiling
IEEE 43-2013 Annex D introduces Insulation Resistance Profiling (IRP) — plotting the full IR versus time curve in discrete increments (typically every 5 seconds) over the full 10-minute test. The shape of the curve, rather than just two points on it, can provide additional diagnostic information. This is particularly relevant when IR₁ exceeds 5000 MΩ and the PI loses reliability.
At the time the standard was published, insufficient data existed to define profile shapes for specific defect types. IRP remains an active area of development and is expected to be addressed in future revisions of the standard.
Documentation
IEEE 43 Section 7 requires the following recorded at every test:
- Ambient temperature, relative humidity, dew point
- Winding temperature
- Time out of service
- Test voltage applied
- Connection arrangement
- IR₁ reading (and IR₁₀ if PI test was run)
- IR₁ corrected to 40°C
- PI value (if applicable)
Keep records for the life of the machine. A single reading in isolation has limited value. A history of corrected readings under consistent conditions is a genuine diagnostic tool.
Quick Reference
| Parameter | Value |
|---|---|
| Standard | IEEE 43-2013 |
| Scope | AC, DC, synchronous machines ≥ 750 W |
| Test polarity | Negative DC |
| Standard test duration | 60 s (IR₁) |
| PI test duration | 600 s → IR₁₀ ÷ IR₁ |
| Reference temperature | 40°C |
| Min IR₁ — pre-1970 windings and field windings | kV + 1 MΩ |
| Min IR₁ — modern AC form-wound (post-1970) | 100 MΩ |
| Min IR₁ — random-wound, form-wound <1 kV, DC armatures | 5 MΩ |
| Min PI — Class A insulation | 1.5 |
| Min PI — Class B and above | 2.0 |
| PI not applicable when | IR₁ > 5000 MΩ; non-insulated field windings; exposed commutators |
| Post-test discharge time | Minimum 4× test duration |
| Temperature correction — IR₁ | Required before comparison (use correct curve for insulation type) |
| Temperature correction — PI | Not required; K_T cancels in ratio |