How to Verify Three-Phase Voltage Balance at a VFD
A variable frequency drive that keeps faulting on input phase loss, undervoltage, or DC-bus ripple is often telling you about its supply, not about itself. The same is true of a motor that runs hotter than its load justifies. Verifying three-phase voltage balance at the drive input is the measurement that separates a supply problem from a drive problem, and it belongs in every drive commissioning and every recurring-fault investigation. This page walks through how to take the measurement, how to compute the unbalance, and how to localize its cause.
Verify Three-Phase Balance at a VFD in one line: To verify three-phase voltage balance at a VFD, measure the three phase-to-phase voltages at the drive input terminals with a true-RMS meter under normal load, average the three readings, and divide the largest deviation from that average by the average itself - that ratio is the voltage unbalance. Repeat the same measurement upstream at the MCC or distribution bus: unbalance present at both points comes from the supply, while unbalance that appears only at the drive points to a local connection problem. Opening an energized enclosure and probing live terminals is qualified-person work under the site electrical safety program; everyone else starts with the drive's own input and bus diagnostics on the display.
What You Need Before Opening the Enclosure
The instrument matters. Drive rectifiers and nearby switching loads distort the voltage waveform, and an averaging meter can report misleading values on distorted power; a true-RMS meter with a measurement category rating appropriate for the point you are probing is the right tool. You also need the site's permission structure: measuring live power terminals is energized electrical work, and whether it may be done at all, with what PPE, and by whom is governed by the site electrical safety program and NFPA 70E-style rules, not by this page. If you are not the qualified person, your job is to request the measurement, not to take it.
Before anyone opens a door, collect the free evidence. Most drives display their measured input or DC-bus voltage in a monitor or diagnostic screen, log undervoltage and phase-loss events with timestamps, and expose those values over the drive's fieldbus. A fault history that clusters at the same time of day, or whenever a particular large load starts, is diagnostic gold and costs nothing. If the drive's diagnostics are polled into a SCADA system, pull the trend before scheduling any live measurement - it often narrows the cause on its own.
Measuring and Computing the Unbalance
The measurement itself is three readings taken phase-to-phase, not phase-to-neutral: L1 to L2, L2 to L3, and L3 to L1, at the drive input terminals, with the drive running its normal load. Phase-to-neutral readings mix in neutral and grounding effects and are not what the motor or the rectifier sees. Take the three readings as close together in time as practical, because loads shift and a reading taken minutes apart is not a fair comparison.
The arithmetic follows the convention used by motor standards: average the three readings, find the reading that deviates most from that average, and divide that deviation by the average. The result, expressed as a percentage, is the voltage unbalance. The reason this small number matters is that a motor - and a drive rectifier - responds to unbalance far more strongly than the number suggests: a small voltage unbalance drives a disproportionately large unbalance in the phase currents, and the extra current shows up as heat. NEMA MG 1 guidance has motors derated when operated above roughly one percent voltage unbalance and advises against operation beyond about five percent, which tells you how little headroom there is.
While the meter is out, the same comparison is worth making on the three phase currents with a clamp meter, because current unbalance is where the damage actually happens. A drive with a healthy, balanced input voltage but strongly unbalanced input currents can indicate a failing rectifier leg or a weak connection, which is a different repair than a supply problem.
Localizing the Cause
One measurement point tells you there is unbalance; two tell you where it lives. Repeat the same three phase-to-phase readings upstream, at the MCC bus or the panel feeding the drive. If the unbalance is already present upstream, the cause is the supply: single-phase loads distributed unevenly across the phases, a utility issue, or a transformer problem, and the fix belongs to whoever owns that level of the distribution. If the upstream bus is balanced and the drive terminals are not, the cause sits between them: a loose or corroded lug, a failing contactor pole, a fuse holder with a bad clip, or a damaged conductor.
Local causes usually announce themselves thermally before they announce themselves electrically. A high-resistance connection dissipates heat, so an infrared scan of the drive's supply path - terminals, disconnect, contactor, fuse holders - taken by someone qualified to open the enclosure will often find the bad joint directly. The extreme case is single-phasing, where one supply phase is lost entirely: the drive typically rides through briefly on its DC bus, then trips on phase loss or bus ripple, and the voltage readings show one pair collapsed relative to the others.
Verifying the Result and Common Mistakes
After any correction - a re-torqued lug per the manufacturer's specification, a replaced contactor, a rebalanced panel schedule - repeat the measurement under a comparable load and confirm the three readings have converged. Then let time finish the verification: if the drive's input diagnostics are trended in a monitoring platform such as Merobix, a week of data showing the phase-loss and undervoltage events gone is the evidence that the repair held, and a recurrence timestamps itself for the next investigation.
The recurring mistakes are worth naming. Measuring phase-to-neutral instead of phase-to-phase produces numbers that do not match the convention the standards use. Measuring at no load hides unbalance that only appears when current flows through a weak joint. Using a non-true-RMS meter on drive-distorted power skews the readings. And blaming the drive for a supply-side unbalance leads to a pointless drive swap; the two-point measurement exists precisely to prevent that.
Frequently Asked Questions
How much voltage unbalance is acceptable for a motor or VFD?
Motor guidance in NEMA MG 1 calls for derating a motor operated above roughly one percent voltage unbalance and advises against running beyond about five percent, and drive manufacturers publish their own input limits in the installation manual. The practical takeaway is that unbalance a meter can barely see is already large enough to matter thermally, so any measurable unbalance that has appeared recently deserves a cause investigation rather than tolerance.
Why does a small voltage unbalance overheat a motor so much?
An unbalanced three-phase supply contains a negative-sequence component that drives current against the motor's direction of rotation. That component sees a very low impedance, so a small voltage unbalance produces a disproportionately large current unbalance, and the extra current heats the windings without producing useful torque. The heating rises much faster than the unbalance number itself, which is why standards derate motors at unbalance levels that sound trivially small.
Can I use the drive's own display instead of a meter?
The drive's monitor values and fault history are the right first step, and they are free and safe to read. They tell you what the drive experienced and when. But most drives report a single input or bus figure rather than three per-phase voltages, so confirming and quantifying unbalance still takes a true-RMS meter across each phase pair, taken by a qualified person under the site's energized-work rules.
Automation services
Need help turning this into a working system?
Merobix integrates SCADA, programs Allen-Bradley and Siemens PLCs, and designs and fabricates industrial control panels.
Meeting requests are reviewed before confirmation.