What a Resistance Reading Tells You and What It Hides

Why this matters

"It ohms out good" is the sentence that precedes more repeat visits than any other in electrical work. It is usually true and almost always incomplete, because the reading it describes was taken under conditions the equipment never operates in: a fraction of a volt, a whisper of current, the machine cold and standing still.

The useful skill is not taking the reading. It is knowing the specific list of faults that reading cannot contain, so that "it ohms out good" becomes "it ohms out good, which rules out these three things and none of the others."

What the instrument is actually doing

A handheld meter in resistance mode puts a small known current through the path and measures the voltage that develops across it. The test voltage is typically a few volts at most, and the test current is somewhere in the microamp to low milliamp range depending on the range selected. Then it divides.

Everything below follows from that one design choice. The meter is asking "what does this path do at approximately zero volts and approximately zero current, right now, at this temperature, standing still?" Every answer it gives is scoped to that question.

Before any of it: resistance is a de-energized measurement, and putting an ohmmeter on a live circuit destroys the meter and puts the resulting arc at the probe tips, inches from your face. Isolate under 29 CFR 1910.333(b)(2) (29 CFR 1926.417 where the work falls under construction), lock and tag, and prove dead live-dead-live per NFPA 70E-2021, 120.5 before the function switch moves to ohms. Establishing that condition includes releasing stored energy: run and start capacitors and drive bus capacitors hold a lethal charge after disconnection, so allow the discharge time stated on the equipment, verify at the terminals with a CAT-rated meter, and never bridge a capacitor's terminals with a screwdriver, which welds the tip and throws molten metal.

What it cannot see, in order of how often it matters

Insulation that only fails at voltage. This is the big one. A winding-to-frame path that a 3 V meter reports as over-range can conduct at 230 V, and conduct hard at the several-hundred-volt peaks an inverter produces. Insulation breakdown is a voltage phenomenon and a handheld meter does not apply voltage in any meaningful sense. That is the entire reason insulation resistance testers exist and apply 250, 500 or 1000 V. IEEE Std 43 governs this test for rotating machinery, sets minimum acceptable values by machine class rather than one universal number, corrects the reading to a reference temperature, and defines the polarization index as the ten-minute reading divided by the one-minute reading with minimum values that depend on the insulation's thermal class. The old field rule of one megohm plus one megohm per kV of rating is a rough floor, not a pass.

Resistance too small for the instrument to resolve. Handheld meters commonly display 0.1 ohm resolution and their own leads contribute a few tenths of an ohm. Measure a 0.05-ohm winding with 0.35-ohm leads and the display reads 0.40 ohms, which is eight times the true value and a 700 percent error contributed entirely by the leads. Null or zero the leads first, and for anything genuinely low-resistance - a bond, a lug, a shunt, a large winding - the instrument for the job is a four-wire (Kelvin) low-resistance meter that drives current through one pair of leads and senses voltage through another, so the lead resistance never enters the answer.

Temperature, which moves the number by a fifth. Copper's resistance changes about 0.393 percent per degree C near room temperature. A winding that reads 4.20 ohms at 22 C will read 4.20 x (1 + 0.00393 x 53) = 5.07 ohms at 75 C, which is 20.8 percent higher with nothing whatever wrong with it. Compare a hot reading against a cold specification and you have condemned a good winding; compare a cold reading against a hot baseline and you have cleared a failing one. Correct one to the other's temperature before you compare them, or state on the ticket that the comparison is uncorrected. Both numbers can be individually right and the comparison still wrong, which is what makes this error survive review.

The parallel path you did not lift. In circuit, the meter reads everything connected across the two probes. A 10.0 k sensor with a 15.0 k resistor across it in the circuit reads (10.0 x 15.0) / 25.0 = 6.0 k, which against a 10.0 k specification looks 40 percent low and reads as a failed sensor. Lift one end of the component before you judge it. This is also why resistance readings across circuit-board components are close to meaningless without the schematic, and why the meter's diode function exists: the test voltage in resistance mode may or may not forward-bias a junction, so the same semiconductor gives different answers on different ranges.

Anything that only exists in motion or at heat. A static reading describes a cold assembly at rest. A terminal that opens under vibration, a conductor broken inside its insulation that makes contact until thermal expansion pulls it apart, a contact that arcs only under load current: all of these read perfectly. The instrument for those faults is a voltage drop taken across the suspect element while the circuit carries its normal current, which a sibling article covers in full.

What the reading is genuinely good for

The exclusions above are not an argument against the instrument. They are an argument for using it on the questions it can answer, and it answers several very well.

• Is this de-energized path continuous? Fuse, switch, contact, conductor, coil. A yes-or-no that no other instrument gives faster.
• Do these identical things match each other? Three windings of a three-phase motor should read within a few percent of one another, measured in one sitting at one temperature. Comparative resistance is far more trustworthy than absolute resistance, because the two biggest error sources, leads and temperature, cancel.
• Does this component match its published curve? Thermistors, RTDs and sensor elements have real published resistance-versus-temperature tables, and a reading taken with the element's temperature actually known is a legitimate calibration check.
• Has this winding shorted turns? A resistance that has fallen well below its own history or below its matching siblings is real evidence, in a way that a resistance that is a little high is not.
• Is this conductor really lifted? Confirming your own isolation before an insulation test.

The worked case: three readings that all said good

A three-phase blower motor that trips its ground-fault protection intermittently, more often on humid mornings, and that the previous tech tested and cleared.

The readings that cleared it. Winding to winding, cold, handheld meter, leads nulled: 4.2, 4.2 and 4.3 ohms. Those match each other within about 2 percent, which is exactly what healthy windings do. Winding to frame: over-range on every combination. Three readings, all normal, and on that basis the motor was returned to service twice.

Why they proved nothing about the complaint. The fault is a path to frame that conducts at line voltage. The handheld meter interrogated it at a few volts and correctly reported that at a few volts nothing conducts. The instrument answered its own question honestly and the tech asked it the wrong question.

The reading that found it. Motor disconnected from its drive and controls first, because insulation test voltage destroys semiconductors and any drive, board or electronic module left connected will be damaged by the test rather than tested by it. Machine isolated, locked and tagged under 29 CFR 1910.333(b)(2), everyone clear of the terminals, because the tester deliberately applies hundreds of volts to a winding you are standing next to. A 500 V insulation test winding-to-frame read 0.8 megohms at one minute and 0.9 megohms at ten minutes.

Two findings there. The absolute value is below even the old one-megohm-per-kV-plus-one rough floor for a machine of this rating, so the path to frame is real. And the polarization index is 0.9 / 0.8 = 1.13, a curve that barely rises, which is the signature of moisture and conductive contamination rather than of thermally aged insulation. That also explains the humid-morning pattern the customer described and the previous tech discounted.

Confirmation. After a controlled dry-out, retested at the same 500 V with the winding at a documented temperature: the one-minute value rose by more than an order of magnitude and the ten-minute-to-one-minute ratio rose with it, which is the pair of changes moisture removal produces. Both values were recorded with their temperature, because the next comparison in two years is only worth making against a reading that carries its conditions.

Verifying a resistance reading before you trust it

Four checks, in this order, each of which takes seconds:

1. Null the leads on the range you are about to use, then short the probes and confirm the display reads zero. A meter that will not zero has a lead fault, and every reading it gives afterwards is high by an unknown amount.
2. Record the temperature with the value. Not "warm" - a number, from a non-contact thermometer. A resistance without a temperature cannot be compared to anything later.
3. Confirm the component is electrically isolated at both ends, not just switched off. If you did not lift it, you measured the circuit, not the part.
4. Take a matching reading on a sibling wherever one exists - the other winding, the other pole, the identical unit next to it - and lead with that comparison rather than with the absolute number.

The failure mode this prevents is subtle and common: a reading that is genuinely accurate, genuinely recorded, and answers a question nobody asked. Write down what the reading rules out, not just what it says.

References

• IEEE Std 43, recommended practice for testing insulation resistance of rotating machinery, including polarization index and temperature correction
• 29 CFR 1910.333(b)(2) with 29 CFR 1926.417 (construction) for de-energizing, lockout and tagging before de-energized testing
• NFPA 70E-2021, 120.5, for the live-dead-live verification sequence and release of stored energy
• Manufacturer documentation for sensor resistance-versus-temperature tables and for disconnection requirements before insulation testing
• See related: How to Use Voltage Drop to Find a Bad Connection; Ohm's Law as a Field Tool Rather Than a Formula