A megohmmeter reading only means something once you know the winding temperature that went with it. The same motor can read 800 MΩ cold in the morning and 90 MΩ after a shift at full load, and nothing is wrong with it. Correcting to a fixed reference temperature is what makes today’s number comparable to last year’s number.
The calculator below applies the IEEE 43 correction, shows the factor it used, and compares the corrected value against the minimum acceptable resistance for your machine type.
How to run a reading through it
- Enter the resistance the megohmmeter showed at the end of the test, in whatever unit the instrument displayed.
- Enter the winding temperature at that moment, not the ambient air and not the nameplate rise.
- Leave the reference at 40 °C unless you keep your records at 20 °C.
- Pick the winding type so the corrected value gets checked against the right minimum.
The factor appears between the two readouts. If it is doing more work than the measurement itself, the flags below the readout will say so.
The math
Insulation resistance falls as temperature rises. The rule of thumb built into IEEE 43 is that it roughly halves for every 10 °C increase, which gives a correction factor:
K = 0.5^((40 − t) / 10)
Multiply the measured value by K and you get the value the machine would have shown at 40 °C.
R₄₀ = R_t × K
Below 40 °C, K is less than 1 and the corrected value drops. Above 40 °C, K is greater than 1 and the corrected value climbs. Cold readings always look better than the machine really is.
Worked example, cold winding
A 6.6 kV motor with form-wound stator coils reads 300 MΩ at 22 °C.
- K = 0.5^((40 − 22) / 10) = 0.5^1.8 = 0.287
- R₄₀ = 300 × 0.287 = 86 MΩ
The raw number looked comfortable. Corrected, it sits under the 100 MΩ minimum for that winding type. That flip is the whole reason the correction exists.
Worked example, hot winding
The same motor is tested right after shutdown and reads 40 MΩ at 65 °C.
- K = 0.5^((40 − 65) / 10) = 0.5^−2.5 = 5.66
- R₄₀ = 40 × 5.66 = 226 MΩ
A number that would have triggered a callout is actually fine.
Where the correction stops being trustworthy
The halving rule is an approximation, and it was drawn from older asphaltic mica and shellac insulation. Treat it as a guide, not a measurement.
Modern thermoset windings are much less temperature sensitive. Epoxy mica insulation in good condition can change far less than a factor of two per 10 °C. Apply the standard factor to a clean thermoset stator tested cold and you can under-report the machine by a wide margin. If you have several years of readings on the same machine at different temperatures, you can derive its own coefficient and use that instead. That is what the alternative factors in the tool are for.
Distance from the reference amplifies the error. At 22 °C the factor is 0.29. At 5 °C it is 0.088. Once the factor is doing more than the meter is, the corrected value is mostly arithmetic. Test near 40 °C where the machine allows it.
Below the dew point, nothing is valid. Condensation on the end windings creates a surface leakage path that has nothing to do with the insulation itself. No correction factor fixes that. Warm the winding above the dew point first, or note the reading as indicative only.
Wet or contaminated windings do not follow the curve. Moisture changes the temperature behavior itself, so a machine that needs the correction most is the one where it works worst. A sharply falling trend across several tests tells you more than any single corrected number.
Which temperature to use
The correction is only as good as the temperature that feeds it.
| Method | When it works | Watch out for |
|---|---|---|
| Embedded RTDs or thermocouples | Machines with installed instrumentation | Slot RTDs read the core, which lags the copper after shutdown |
| Winding resistance measurement | Most accurate for the copper itself | Needs a known cold resistance and reference temperature |
| Infrared on the end windings | Quick check on an open machine | Surface only, and emissivity of varnish varies |
| Ambient air after long shutdown | Machine off and settled for many hours | Only valid once the whole mass has equalized |
A machine that ran until an hour ago has a temperature gradient in it. Wait for it to settle or accept that the correction is approximate.
IEEE 43 minimum values at 40 °C
The corrected one-minute reading gets compared against these.
| Minimum IR (1 min, 40 °C) | Applies to |
|---|---|
| 5 MΩ | Machines with random-wound stator coils, and form-wound coils rated below 1 kV |
| 100 MΩ | Most DC armature windings and AC stator windings with form-wound coils built after about 1970 |
| kV + 1 MΩ | Most windings built before about 1970, all field windings, and anything not covered above |
kV is the rated machine terminal-to-terminal voltage, rms.
These are floors for deciding whether a machine is safe to energize or overpotential test. They are not condition targets. A healthy 6.6 kV stator that has always read 2 GΩ and now reads 150 MΩ passes the table and still needs looking at.
FAQ
Do I correct before or after calculating the polarization index?
Neither, in practice. PI is a ratio of two readings taken ten minutes apart on the same winding at roughly the same temperature, so the correction factor cancels out. Correct the one-minute value if you want to log it, but do not correct the two readings separately and then divide.
My meter has an automatic temperature correction feature. Should I use it?
You can, but log the raw reading and the temperature as well. Instruments do not all use the same coefficient, and you cannot re-derive the measurement later from a corrected number alone.
What if I test at exactly 40 °C?
The factor is 1 and the reading stands as measured. That is the point of testing near the reference.
Can I use this for cables, transformers, or switchgear?
The arithmetic is the same, but the coefficient is not. IEEE 43 covers rotating machinery. Cable and transformer insulation systems have their own correction tables, and many utilities normalize cables to 20 °C rather than 40 °C. Use the 20 °C reference option and a coefficient from the relevant standard or the manufacturer.
Why is my corrected value so much lower than the raw reading?
Because you tested cold. A reading at 15 °C gets multiplied by about 0.18. That is expected, and it is why comparing an uncorrected winter reading against an uncorrected summer reading tells you nothing.
Should I keep the raw or the corrected value in the test record?
Both, plus the temperature and the coefficient used. If any of the three is missing, the record cannot be re-checked later.