Troubleshooting
Setting voltage imbalance limits on a 480 V phase monitor
Short answer
Keep voltage unbalance at the motor terminals at 1% or less to run full rated power; above that NEMA MG 1 requires derating, reaching 75% of rated power at 5%, with operation above 5% not recommended. A monitor set to 15% voltage imbalance is far too loose to protect a motor. Measure A-B, B-C and C-A at the starter stopped and running, and at the main panel, and call the utility only when the main panel itself is unbalanced.
On this page
- What is the acceptable voltage imbalance on a 480 V three-phase motor?
- How do you calculate three-phase voltage unbalance from panel readings?
- Are 15% voltage and 10% phase imbalance settings too loose for motor protection?
- Why does a small voltage imbalance overheat a motor?
- Where do you measure to separate a utility fault from a facility fault?
- When should you call the utility about voltage imbalance?
- What protection stops single phasing from burning up the motor?
- How do you verify the repair?
- Questions asked
Keep voltage unbalance at the motor terminals at 1% or less: that is where the motor delivers its full rated power. Above 1%, NEMA MG 1 requires derating, and at 5% unbalance the motor is down to 75% of rated power, with operation above 5% not recommended at all. A phase monitor set to a 15% voltage-imbalance tolerance is set about three times looser than the level where the standard stops recommending operation, so the settings are the first thing to change. Two motors already lost to phase loss point at a supply or protection problem, and the measurements below separate your wiring from the utility's.
What is the acceptable voltage imbalance on a 480 V three-phase motor? #
One percent voltage unbalance or less lets the motor deliver full rated power; NEMA MG 1 requires derating above 1%, allows up to 5% only at 75% of rated power, and does not recommend operation above 5% at all.
Those numbers come from NEMA MG 1, Motors and Generators, the standard that defines AC motor performance limits; the medium-motor derating factor for unbalanced voltage is Part 14 of the 2016 edition (Revision 1, 2018) and appears as Figure 20-2 in the 2011 edition. Guidance published with the standard also notes that most motor warranties stop at 1% unbalance. Confirm the motor's own documentation and any local requirements with the authority having jurisdiction before you lock in settings.
| Voltage unbalance | Deviation from the average | Consequence for the motor |
|---|---|---|
| 1% or less | 4.8 V or less | Full rated power |
| 2% | 9.6 V | Derate per the NEMA MG 1 curve |
| 5% | 24 V | 75% of rated power |
| Above 5% | Above 24 V | Operation not recommended |
| 15% (the current setting) | 72 V | Three times the not-recommended level |
The deviations in the middle column assume a line-to-line average that lands exactly on the system's nominal 480 V; base every number on the measured average of your three readings instead.
How do you calculate three-phase voltage unbalance from panel readings? #
Divide the largest deviation of any one line-to-line voltage from the three-reading average by that average and multiply by 100; percent unbalance = maximum deviation ÷ average × 100, using A-B, B-C and C-A readings, not phase-to-ground.
On an ungrounded or high-resistance-grounded system at this voltage, phase-to-ground readings wander with ground-detector leakage and will mislead you; unbalance is defined on line-to-line voltage. Worked example, using illustrative readings on the fan starter:
- Stopped, at the starter: A-B 479 V, B-C 480 V, C-A 478 V. Average = (479 + 480 + 478) ÷ 3 = 479.0 V. Maximum deviation = 1.0 V. Unbalance = 1.0 ÷ 479.0 × 100 = 0.21%.
- Running, at the starter: A-B 468 V, B-C 479 V, C-A 480 V. Average = (468 + 479 + 480) ÷ 3 = 475.7 V. Maximum deviation = 475.7 − 468 = 7.7 V. Unbalance = 7.7 ÷ 475.7 × 100 = 1.6%.
- Running, at the main panel: A-B 477 V, B-C 478 V, C-A 477 V. Average = 477.3 V. Maximum deviation = 0.7 V. Unbalance = 0.7 ÷ 477.3 × 100 = 0.14%.
The running starter reading fails the 1% limit. Interpolating linearly between the standard's documented points, factor 1.00 at 1% and 0.75 at 5%, gives 1.00 − (1.6 − 1) ÷ (5 − 1) × 0.25 ≈ 0.96 of rated power; that interpolation is an estimate and the published curve governs. Practically, you find and fix the connection rather than plan to run the fan derated. The pattern across the three locations — balanced stopped, unbalanced running, balanced at the panel — is what localizes the fault in the next sections.
Are 15% voltage and 10% phase imbalance settings too loose for motor protection? #
Yes: 15% voltage imbalance tolerates a 72 V deviation at a 480 V average, three times the 5% level where NEMA MG 1 no longer recommends running the motor at all, so those settings will not protect it.
On this monitor the entered tolerances are 15% voltage imbalance and 10% phase imbalance. Those are settings, not motor limits: 10% means a 48 V deviation and 15% means 72 V before the relay reacts, while the motor starts losing capacity past a 4.8 V deviation at this average. Check the monitor's instruction sheet for what each setting actually computes, because "voltage imbalance" and "phase imbalance" are not always the same formula in one relay. Then set both from the motor side of the argument:
- Voltage-unbalance trip: 2%, about a 9.6 V deviation, with a delay of a few seconds so the starting voltage dip does not nuisance-trip it.
- Warning or alarm level: 1%, the full-power limit.
- Phase-loss and undervoltage functions: leave them active; a lost phase must trip, never just warn.
Do not raise a setting to silence warnings, and do not bypass the trip output to keep the fan running.
Why does a small voltage imbalance overheat a motor? #
Unbalanced voltage drives negative-sequence current through the rotor; motor current unbalance typically runs 6 to 10 times the voltage unbalance, so the 1.6% in the example becomes roughly 10 to 16% on the current.
Negative-sequence current rotates against the rotor's field and dissipates as heat the frame cannot shed, so the rotor and stator cook unevenly while the average line current still looks normal. That is how a fan burns up on voltages that never trip anything, the failure mode behind why motors burn up with no overload trips on 480 V. Measure the three motor line currents whenever you measure the voltages and judge the spread against nameplate full-load current, as covered in three-phase motor current unbalance: acceptable ranges and analysis.
Where do you measure to separate a utility fault from a facility fault? #
Measure A-B, B-C and C-A at the starter with the motor stopped, repeat with it running, then repeat at the main panel: whichever location first shows the unbalance sits just upstream of the fault.
These are live measurements at this voltage: use a true-RMS meter rated CAT III 600 V or better at the starter (CAT IV at the service entrance) and the PPE and procedures your site requires under NFPA 70E, 2024 edition, which covers electrical work practices for energized equipment. When a reading sends you toward inspection or termination work, stop, lock out, de-energize and test for absence of voltage before opening anything.
- Record the monitor's displayed fault, the event time and the load state before any reset.
- At the starter input, measure A-B, B-C and C-A with the motor stopped; write down all three.
- Start the fan and repeat the three readings under its normal load.
- Measure the three motor line currents while running and note the spread.
- Repeat the three voltages at the main panel under the same load.
- Calculate percent unbalance at each point from the measured average, as shown above.
- Inspect, de-energized, the path between the last balanced point and the first unbalanced one.
| Pattern in the readings | Fault location |
|---|---|
| Main panel unbalanced under normal load | Utility supply, service conductors or upstream transformer: document and call |
| Main panel balanced, starter unbalanced stopped and running | Facility feeder, disconnect, fuses, contactor poles, terminations |
| Starter balanced stopped, unbalanced running | High-resistance connection exposed by load current |
| Voltages balanced, motor currents badly unbalanced | Motor windings, motor leads, the load, or the measurement itself |
| One phase missing or two line-to-line readings collapsed | Open fuse, contactor pole, conductor or connection |
The worked example is the third row: 0.21% stopped becomes 1.6% running at the starter while the main panel holds 0.14%, which puts the defect on the facility side of the panel. Compare voltage across each device while loaded; a large drop across one pole or fuse identifies the resistance in that path. If the same current unbalance appears on every motor on the service, the cause is upstream of all of them, the situation covered in diagnosing current imbalance across every motor on a service.
When should you call the utility about voltage imbalance? #
Call when the three main-panel phase-to-phase voltages, measured under normal load and again with large single-phase loads off, hold above 1% unbalance on repeated timed readings; the logged warning count alone is not evidence.
Those 25 entries in the trend log are a frequency count: they carry no magnitude, duration, load state or location, and a monitor set to 15% tolerance logs only extreme events anyway. What moves a utility is a documented pattern. Take the three main-panel voltages at two or three times of day under normal load and once with your big single-phase loads off, because your own distribution can unbalance the service. If the main panel still exceeds 1% unbalance on repeated readings after the feeder and loads check out, call with the numbers, timestamps and load state. Utilities do adjust transformer taps and rebalance single-phase loading when handed data, and they expect you to have cleared your own wiring first.
What protection stops single phasing from burning up the motor? #
A solid-state overload relay with phase-loss detection trips within seconds of losing a phase, and the phase monitor's trip output should drop the starter coil and stay latched until reset; you need both working together.
Two burned motors say the protection never acted: the loose monitor settings watched the supply fail while nothing modeled the motor. Set the overload from the motor nameplate full-load current, and prefer a solid-state overload with phase-loss detection for this duty. Current-based protection still has blind spots, detailed in what current-only motor protection can and cannot detect, which is why the phase monitor stays in the scheme once its settings match the motor. Wire the monitor's trip output in series with the starter coil so any trip removes the run command, and keep the latched-off behavior until an operator deliberately resets it: automatic restart when the supply recovers is a machine hazard, not a convenience.
How do you verify the repair? #
Re-measure the three voltages at the starter stopped and running and recalculate percent unbalance; you want 1% or less under load, three motor currents within a few percent of each other, and no new monitor events over a full operating cycle.
After the repair, repeat the stopped and running voltage sets at the starter and at the main panel and check the motor currents against the nameplate and the load. Run the fan through its normal duty cycle and read the monitor's trend log for new events. Prove the trip path with the monitor's own test or simulation function from its instruction sheet, never by opening a fuse or conductor on a running circuit, and confirm that a real trip drops the starter and that the reset restores it exactly once, deliberately.
Questions asked
Does the 15% imbalance setting on my monitor harm the motor?
The setting creates no unbalance itself, but it hides it. Until you tighten it, the monitor is a recorder, not protection: at the line-to-line average it stays quiet while one line-to-line voltage sits 72 V off the average, well past the 5% level where motors should not run even derated. The motor absorbs every event those logged warnings counted.
How many imbalance warnings in the trend log justify calling the utility?
None by itself. The counter records how often something happened, not how big it was, how long it lasted or where it came from. The call is justified when you can hand over timed three-voltage readings from the main panel above 1% unbalance after your feeder checks clean, and saying you already ruled out facility wiring gets a faster response.
Why is the voltage balanced with the fan off but unbalanced when it runs?
A high-resistance joint, such as a loose lug, a corroded fuse clip or a pitted contactor pole, drops voltage only when current flows through it. With the motor off, almost no current flows and the fault is invisible on a voltmeter. Always take the second set of A-B, B-C and C-A readings with the fan running under normal load.
Can the phase monitor replace the overload relay?
No. The monitor watches supply voltage at its terminals; the overload relay models motor heating from current. A solid-state overload with phase-loss detection trips within seconds of a lost phase, which is what cooked the previous motors, while a thermal unit reacting to average heat can be slow on a lightly loaded fan. Size it from the nameplate full-load current and keep both devices.
What does 1% voltage unbalance equal in volts at a 480 V system?
About 4.8 V: the most any one line-to-line reading may deviate from the three-reading average when that average is exactly 480 V. Use the measured average, not the nominal voltage, as the base. At a 476 V measured average, the 1% limit is a 4.76 V deviation.