The Motor That Ran Hot on a Healthy-Looking Supply

Why this matters

Two motors on the same pump had been replaced in fourteen months. Both times the tech on site did competent work: megger readings on the new motor were good, the coupling was aligned, the overloads matched the nameplate, and the unit ran when they left. The third failure came in as a warranty argument with the motor supplier. It was not a motor problem, and every piece of evidence needed to say so was already written on the first two tickets. Nobody had done the one piece of arithmetic that turns three ordinary-looking voltage readings into a verdict.

This is that arithmetic, and the reasoning that hangs off it.

Reading the file before touching the equipment

The unit was down and the customer wanted a diagnosis before another motor was ordered, so the first hour went to the service history rather than the roof. Three things were in it.

Ticket one, fourteen months back: "Motor failed, insulation to ground low on two windings. Replaced. Checked voltage, 484 / 462 / 470, within range. Checked amps, one leg slightly high, motor drawing near nameplate."

Ticket two, five months back: "Motor failed same as before. New motor megged good. Overloads did not trip. Replaced motor, ran unit, left running." No electrical readings recorded.

Ticket three, the current call: same symptom, roughly the same interval, and a customer who has stopped believing the motor is the problem.

The first ticket is the whole case. Somebody wrote down three line-to-line voltages, looked at them one at a time against the equipment's listed range, found all three acceptable, and moved on. Read one at a time they are fine. Read against each other they are not, and a motor does not experience them one at a time.

The number nobody calculated

Percent voltage unbalance, as NEMA MG 1 defines it for the motor manufacturers that build to it, is the maximum deviation of any one line-to-line voltage from the average of the three, divided by that average. It is not the spread between highest and lowest, and it is not a percentage of nominal.

Run the first ticket's readings through it.

• Average: 484 plus 462 plus 470 is 1416, divided by 3 is 472 V.
• Deviations from that average: 484 is 12 V high, 462 is 10 V low, 470 is 2 V low.
• Largest deviation is 12 V. 12 divided by 472 is 0.0254, so 2.5 percent unbalance.

Every individual reading sat inside the plus or minus 10 percent band most utilization equipment is listed for, which is why the tech passed it. Unbalance is a separate test with a far tighter limit, and it is the test this motor was failing.

Why a small voltage unbalance is a large current unbalance

This is the mechanism, and it is worth carrying because it explains the ratio rather than asking you to memorise it.

An unbalanced three-phase set decomposes into a balanced positive-sequence set that rotates the way the motor turns, and a balanced negative-sequence set that rotates the opposite way. The motor presents very different impedances to the two. To positive sequence at running slip it looks like a normally-loaded motor. To negative sequence, the rotor is effectively moving backwards through that field at nearly twice synchronous speed, so the motor presents something close to its LOCKED-ROTOR impedance instead.

That is why the multiplier is what it is. A small negative-sequence voltage pushed through an impedance several times lower than the running impedance produces a proportionally larger current. The multiplier is approximately the motor's own locked-rotor-to-full-load current ratio, which is why it varies by machine and why the commonly quoted range is roughly 6 to 10 times. It is a consequence of that motor's design, not a universal constant, and the motor's own data governs when you have it.

The heating consequence follows the same route. Negative-sequence current induces rotor currents at close to double line frequency, in a rotor built for slip frequency, so the loss lands where the motor is least able to shed it. The approximation reproduced in motor literature is that the temperature RISE increases by roughly twice the square of the percent unbalance, stated for a motor operating at full load. At 2.5 percent that is about 12.5 percent more rise. Read the direction check: at 1 percent it is about 2 percent more rise, which is genuinely nothing, and that is why the same relationship is not a reason to panic over a slightly imperfect supply.

Reconstructing what the currents must have been

The first ticket did not record the three currents, only the observation that one leg was slightly high and the motor was near nameplate. That is enough to reconstruct, because the mechanism fixes the relationship.

Take the motor at 95 percent of its 27.0 A nameplate full-load current, so an average of about 25.7 A across the three legs. At a locked-rotor-to-full-load ratio of 6, a 2.5 percent voltage unbalance produces roughly 15 percent current unbalance. Fifteen percent of 25.7 A is 3.9 A of deviation, so the three legs would read approximately 29.6, 24.0 and 23.5 A. Add them: 77.1 A, divided by 3 is 25.7 A, and the largest deviation is 3.9 A, which is 15.2 percent. The reconstruction closes.

Now read it the way the tech on ticket one did. Two legs are comfortably under nameplate. One leg is 29.6 A against a 27.0 A nameplate, which is about 110 percent, and gets written down as "slightly high." The overload relay has an element on every leg, including the high one, so it is not averaging anything. It does not open because 29.6 A is simply below its trip current: set at the 27.0 A nameplate with the service-factor allowance most relays carry, it is looking for something above about 31 A. So the motor cooks with a protective device watching the exact leg doing the damage and correctly declining to act, because thermal damage from negative-sequence current arrives well below the current an overload is built to find. Worth asking on any repeat failure: many electronic overload relays carry a phase-unbalance or phase-loss function that would have caught this, and it is frequently left disabled. So the motor cooks with no protective device reporting anything, on a supply where every voltage is in range and two of the three currents are low.

That is the entire trap in one sentence: the readings that look reassuring are the two legs that are not doing the damage.

What that unbalance was doing to the motor's own rating

NEMA MG 1 sets out a derating curve for motors on unbalanced voltage. The factor is 1.0 at no unbalance, falls to roughly 0.95 at 2 percent and to roughly 0.75 at 5 percent, and the standard directs that motors not be operated above 5 percent unbalance at all. It binds through the motor manufacturer's own rating rather than on its own authority, so where the motor's literature states a different curve or a tighter limit, the manufacturer's number governs and the curve here is the general shape.

At 2.5 percent, the derating factor sits at roughly 0.93. The motor was carrying about 0.95 of nameplate. The load was therefore slightly above what the supply permitted that motor to carry continuously, before you add the 12.5 percent extra temperature rise from the same unbalance. That combination does not fail a motor in a week. It fails one in a year, which is exactly the interval the tickets show, and it fails it thermally, so the replacement meggers perfectly and the supplier's warranty inspection finds nothing wrong with the machine.

Separating three causes that produce the same reading

Unbalance at a motor has three usual origins and they call for three different fixes, so the field work is about telling them apart rather than about confirming the number again.

Every reading below is taken on energized conductors, which puts it under 29 CFR 1910.333(a)(1): de-energize before working on or near exposed live parts unless the employer can demonstrate that de-energizing introduces additional or increased hazards or is infeasible due to equipment design or operational limitations. Voltage and current readings that only exist under load are the ordinary case for the second clause. Work them under the employer's energized-work program with boundaries and PPE established under NFPA 70E-2021 130.5 and 130.7, in the edition your employer's program or your authority having jurisdiction has adopted. Do not put a hand on a lug to check for heat: read connection temperature with a non-contact infrared instrument from outside the restricted approach boundary, or through an infrared window if the enclosure has one.

Unequal single-phase loading upstream. A 480/277 V system carrying a large block of 277 V lighting or single-phase load landed unevenly across the three phases pulls one line-to-line pair down. The tell is that it moves with a schedule. Take the same three voltages twice, once with the building's single-phase load on and once with it off, and an unbalance that changes materially between the two belongs to load distribution, which is a redistribution job in the panel, not a motor job.

A high-resistance connection between the service and the motor. A loose lug, a corroded splice or a partially backed-out terminal drops voltage on one leg only, and because the drop is current-driven, the unbalance WORSENS as the motor loads and shrinks toward nothing at idle. The tell is that time-of-day makes no difference but load does, and the connection usually shows a temperature difference against its two siblings on an infrared scan taken from outside the boundary.

Utility-side unbalance. Present at the service entrance itself, unchanged by anything inside the building, and it appears identically at every panel. That is a utility call, not a customer repair.

Measure at the service first and at the equipment second. If the unbalance is already at the service, nothing inside the building created it. If it is small at the service and large at the equipment, everything between them is a suspect and the connection tests are worth doing before anyone quotes a new feeder.

The test that says motor or supply

If a motor keeps being blamed, one de-energized change settles it. With the circuit isolated, locked and tagged under 29 CFR 1910.333(b)(2), or 29 CFR 1926.417 on construction, and separately under 29 CFR 1910.147 for the mechanical side, because a pump can be turned by its coupling or driven backwards by static head on the discharge and 1910.147's carve-out at (a)(1)(ii)(C) is of the electrical exposure only rather than of the machine, and proved dead live-dead-live per NFPA 70E-2021 120.5, roll all three motor leads one position, so that the conductor that fed T1 now feeds T2, T2 feeds T3, and T3 feeds T1. Rolling all three preserves the phase sequence, so the motor turns the same direction and no rotation-sensitive equipment downstream is affected. Swapping only two would reverse it, which on a positive-displacement machine is a different and worse problem.

Re-run and read the three currents again. If the high leg follows the SUPPLY conductor to its new terminal, the problem is on the supply side. If the high leg stays on the same MOTOR terminal, the winding or the motor's internal connection owns it. On this pump, the high leg followed the supply conductor, which ended the warranty argument before the third motor was ordered.

Confirming the fix held

Do not close the call on a corrected voltage reading alone, because the voltage recovers the instant the load comes off and the number you want is the loaded one. Take the three line-to-line voltages and the three line currents at the motor with the machine at its normal duty, compute percent unbalance on both sets, and write all eight numbers on the ticket rather than a verdict. Then repeat them at the far end of a duty cycle, an hour or more in, since a marginal connection recovers when cold and reappears warm. A supply that reads under 1 percent voltage unbalance loaded, with current unbalance in single figures, has stopped being the reason a motor dies at fourteen months. Leave the recorded set behind, because the next tech's whole problem this time was that ticket two had none.

References

• NEMA MG 1, motors and generators, for the definition of percent voltage unbalance and the derating curve, as adopted through the motor manufacturer's rating; the specific motor's own literature governs where it differs
• 29 CFR 1910.333(a)(1), general industry electrical safety-related work practices and the de-energizing gate
• NFPA 70E-2021, 120.5, 130.5 and 130.7, as adopted by your employer's electrical safety program or your authority having jurisdiction
• See related: Single-Phase and Three-Phase in Practice; Why Voltage Sags Under Load