Voltage Imbalance on Three-Phase Refrigeration: How to Measure It and When to Shut Down

A three-phase compressor that trips intermittently, with every reading at rest looking perfectly normal, is one of the most commonly misdiagnosed faults in the trade — and it is very often not a compressor fault at all, but a supply imbalance quietly cooking one winding every hour the plant runs. Three voltage measurements and a calculator settle it. By the end of this article you’ll be able to run the NEMA imbalance calculation correctly, know the current-amplification relationship that follows from it, and make the shut-down decision before a second compressor gets replaced for nothing.

Safety note: the voltage and current readings below are taken live, at the isolator or terminals under load. Use a CAT-rated meter and clamp appropriate to the panel, and only isolate, lock off and prove dead per the OHS Act 85 of 1993 when the decision above calls for a repair rather than a measurement.

The NEMA method, defined

Measure all three line-to-line voltages. Average them. Find the single largest deviation of any one reading from that average. Express that deviation as a percentage of the average — that is the NEMA definition of voltage unbalance, and NEMA’s limit is 2%.

% imbalance = (largest deviation from the average) ÷ average × 100

Above 2%, derate the load or don’t run it. This isn’t a conservative guideline with room to negotiate — the effect on the motor is disproportionate to the voltage figure, which is exactly what makes this fault so easy to underestimate at a glance.

Worked example: 3.9% imbalance mistaken for a failing compressor

A three-phase compressor trips intermittently on its overload. Nameplate FLA is 20 A. The customer wants the compressor replaced — it has already been replaced once this year.

Voltage check, all three line-line pairs: 415.6 V, 400.0 V, 384.4 V.

Average = (415.6 + 400.0 + 384.4) ÷ 3 = 400.0 V. The largest deviation is 15.6 V. Voltage imbalance = 15.6 ÷ 400.0 × 100 = 3.9% — already about double the NEMA limit before a single current reading is taken.

Current check, running, stable load: L1 = 14.5 A, L2 = 20.5 A, L3 = 26.5 A. Average = (14.5 + 20.5 + 26.5) ÷ 3 = 20.5 A. Largest deviation = 6.0 A. Current imbalance = 6.0 ÷ 20.5 × 100 = 29.3%.

The 6-10x current amplification rule

A 29.3% current imbalance looks, on its own, like a serious motor problem — well past the 10% guideline for running current balance. But cross-check it against the voltage figure using the amplification relationship: current imbalance in an induction motor typically runs at 6 to 10 times the percentage voltage imbalance that caused it, because the negative-sequence component sees a very low rotor impedance. A small voltage problem at the supply produces a much larger current problem at the motor — which is exactly why voltage balance belongs at the top of the test sequence, checked before a current imbalance gets treated as a motor fault.

Here, 3.9% voltage imbalance predicts a current imbalance of 3.9 × 6 = 23.4% to 3.9 × 10 = 39%. The measured 29.3% sits squarely inside that band — fully explained by the supply imbalance, not evidence of an additional winding fault. Winding resistance across all three pairs comes back within 5% of each other, and insulation resistance reads 240 MΩ on all three, confirming the windings are sound. The fault is upstream: an unbalanced single-phase load on the site, a loose or high-resistance connection, or an imbalance on the incoming supply. A second compressor failure on the same site within a year almost always means an upstream problem was never found the first time.

Reference: imbalance limits and what they mean

As a working figure, roughly 3% voltage imbalance raises winding temperature about 25%, and insulation life halves for every 10°C above rated temperature. That is why the 2% NEMA figure is treated as a hard limit rather than a target to approach.

Phase loss: the acute version of the same fault

Lose one line entirely while a three-phase compressor is running — a blown fuse, a failed contactor pole, a broken conductor — and the motor does not stop. It keeps turning on the remaining two live lines, and to keep delivering anywhere near the same shaft power, current in those two phases rises to roughly 1.73 times its normal value, the √3 factor falling straight out of the power relationship for a three-phase load losing one leg while still trying to do the same work. Windings sized for balanced current, suddenly carrying 73% more through two-thirds of the winding set, overheat within minutes — the fastest way to kill a three-phase compressor electrically. This is the case for a dedicated phase-failure relay on three-phase compressor circuits rather than relying on the thermal overload alone, since it drops the contactor within a second or two of losing a leg, well before winding temperature becomes a problem.

The diagnostic signature is unmistakable: one line reading near 0 V or 0 A, the other two elevated and roughly equal to each other. It is easy to confuse the early symptoms of phase loss with a pitted contactor pole feeding into an overload trip, so trace the fault back to its actual location — fuse, contactor pole, or supply — rather than stopping at the first low reading.

When to shut down: the decision comes before the repair

The shut-down decision is made at the voltage-imbalance calculation, before any repair work begins and before winding tests are even run. A worked example from an actual site makes the sequence clear. Line voltages at the isolator, unit off: L1-L2 398 V, L2-L3 402 V, L1-L3 386 V. Average = 395.3 V. Deviations of 2.7 V, 6.7 V and 9.3 V, maximum 9.3 V. Imbalance = 9.3 ÷ 395.3 × 100 = 2.36% — over the 2% limit even though the average voltage itself, 395 V, sits comfortably within the normal 360-440 V band.

At 2.36% imbalance, expect roughly 14-24% current imbalance and a 15-20% winding temperature rise in the worst-affected phase — enough on its own to explain a repeated overnight overload trip with no fault in the compressor. The decision not to run is made here, before a single repair tool comes out: starting into a 2.36% imbalance would trip it again within the hour, and each restart at four to seven times FLA does more damage than the wait. Line-to-neutral readings then narrowed the fault to a discoloured, loose lug on L1, re-terminated to SANS 10142-1 and torqued correctly. Imbalance afterward measured 0.4%. All isolator work was carried out dead, locked off and proven dead, as required under the OHS Act 85 of 1993.

Load shedding makes this worse, not just more frequent

Every restart after a load-shedding recovery draws locked rotor current, typically four to eight times nameplate FLA, at exactly the moment voltage across the local network is depressed by every machine in the area reconnecting and inrushing together. Repeated restarts in quick succession stack thermal soaks into the windings faster than they dissipate, and a marginal voltage imbalance that a compressor tolerates once becomes a compounding problem across a stage-recovery evening. An anti-short-cycle timer, holding the compressor off for a minimum of three to five minutes after any stop, is cheap insurance against this pattern and is worth specifying on any site with frequent load-shedding cycling — a fuller treatment of the mechanism is in our article on what load shedding does to refrigeration equipment.

Final Thoughts

Three line-to-line voltage readings, an average, and one percentage calculation tell you more about a tripping three-phase compressor than a full winding teardown will. Run the NEMA calculation before condemning any compressor, cross-check current imbalance against the 6-10x relationship before treating it as a motor fault, and make the run-or-shut-down call at the voltage stage rather than after further damage is done. The complete method, with every supporting calculation, is set out in the complete guide to electrical troubleshooting and fault-finding.