Troubleshooting
Motor starters trip when MCC doors are closed
Short answer
The trips are real protection, not nuisance: a thermal overload relay without ambient compensation responds to enclosure air temperature, so doors open plus a fan suppresses trips while motor heating continues. A 10-12 V spread on 480 V is 1.0% to 1.7% voltage unbalance, above the NEMA MG 1 limit of 1.0%, and a 60-70 A current spread is severe. Close the doors, pull the fan, and fix the upstream supply that survived the chiller and MCC replacement.
On this page
The trips are protective operations being suppressed, not nuisance trips to be cooled away. A thermal overload relay without ambient compensation responds to the air around it as well as to current: with the doors closed the enclosure runs hot and the relay trips earlier; with the doors propped open and a fan on the starters, the relay cools and stops tripping while motor heating is unchanged. Meanwhile the reported 10-12 V phase-to-phase spread on a 480 V system works out to 1.0% to 1.7% voltage unbalance, at or above the 1.0% limit in NEMA MG 1, and a 60-70 A current spread is severe. Close the doors, take the fan away, and diagnose the supply: the imbalance survived replacement of both the chiller and its MCC, so it lives upstream of both.
Why the fan stops trips that should happen #
A bimetallic or eutectic thermal overload relay models winding heating with a heater element in each phase; the trip mechanism releases when any element deflects far enough. That deflection has two inputs: the element's own I²R heating from current, and the ambient air in the bucket. An ambient-compensated relay cancels most of the second input; a plain one does not. Hotter enclosure air moves the trip point down, cooler air moves it up, and the fan changes only that term. The current term — set by mechanical load and by phase unbalance — is untouched, which is how a motor can sit in the trip zone all day with a fan blowing on its starter.
Enclosure heat has real sources: bus and starter losses, resistive terminations, adjacent buckets, blocked ventilation and clogged filters. A doors-closed trip therefore means one of two things, both genuine: motor current above what the winding tolerates, or enclosure air hotter than the application assumed. The fan silences either one without fixing either one. The doors themselves are also part of the protection: NFPA 70, 2023 edition, Section 110.27 requires live parts to be guarded by enclosures or other suitable means, and a propped-open 480 V door leaves those parts exposed while defeating the enclosure's ingress and fault-containment functions.
What the spread numbers actually are #
A spread is not a percentage. NEMA MG 1 (2016 edition, revised 2018) computes voltage unbalance as 100 times the maximum deviation of any line-to-line voltage from the average of the three, divided by that average, read simultaneously at the motor terminals under load. Dividing the span by 480 V and comparing the result with the ±10% voltage variation that NEMA MG 1 Section 12.44.1 permits at rated frequency mixes two different quantities: that tolerance is for all three phases moving together, and the 1.0% unbalance limit applies alongside it, not inside it. Editions and local amendments differ; confirm applications with the authority having jurisdiction and the equipment documentation.
The reported spread brackets tightly. A 10 V span gives at least about 1.0% (a 5 V deviation against a 480 V average); a 12 V span gives 1.25% when the middle reading sits at the average, and up to about 1.7% when two readings sit high. Every arrangement of a 10-12 V spread on 480 V is at or above the limit, and above 1.0% NEMA MG 1 requires derating the motor per its derating-factor curve (Figure 20-2). By the MG 1 method, 3% voltage unbalance adds about 18.5% heating, and the working rule that every 10 °C of added winding temperature roughly halves insulation life means that much unbalance can cut winding life by more than half. Repeated tower-fan and pump motor failures and a hermetic compressor burnout are consistent with years of exactly that.
Unbalance that survives replacement of the chiller and its MCC, and that shows on more than one machine, points at the common source. When every motor on one service reads the same imbalance, the fault boundary is upstream of all of them, and the measurement sequence in diagnosing current imbalance across every motor on a service applies directly.
The arithmetic on the reported spreads #
Illustrative inputs matched to the report — line-to-line voltages of 486 V, 480 V and 474 V read simultaneously at a chiller motor's terminals on a 480 V system, and line currents of 158 A, 148 A and 90 A at the starter output:
- Voltage: average = (486 + 480 + 474) V ÷ 3 = 1440 V ÷ 3 = 480.0 V; largest deviation = 486 V − 480.0 V = 6.0 V; unbalance = 6.0 V ÷ 480.0 V × 100 = 1.25%.
- Current: average = (158 + 148 + 90) A ÷ 3 = 396 A ÷ 3 = 132.0 A; largest deviation = 132.0 A − 90.0 A = 42.0 A; unbalance = 42.0 A ÷ 132.0 A × 100 = 31.8%.
The decision on those numbers: 1.25% puts the motor outside its successful-operation region at rated load, and 31.8% current unbalance — well out of proportion to the voltage unbalance — says the motor's own current path (winding condition, internal or field connections) is contributing alongside the supply. Against an illustrative 140 A nameplate, the 158 A phase carries 113% of full-load current continuously while the 132 A average looks healthy, and the relay trips on its hottest element, which is that phase. The trips are genuine; the settings stay, the doors close, and the unbalance gets corrected.
Diagnosis with the doors closed and latched #
Before any door or bucket is opened, establish an electrically safe work condition — de-energize, lock out, test for absence of voltage — as NFPA 70E, 2024 edition, requires for that work. The running measurements below happen with the doors closed and latched.
- End the workaround as an operating condition: doors closed and latched, fan removed, trips expected and logged rather than prevented.
- From the nameplates, record motor voltage, full-load current and service factor, and verify each overload heater or dial against the relay manufacturer's selection table for that current — the same nameplate-first sequence as selecting overload and jam protection for a reversible motor. NFPA 70, 2023 edition, Article 430.32 caps separate overload protection at 125% of full-load current for a 1.15-service-factor motor, 115% otherwise.
- With the chiller at normal load, read the three line-to-line voltages simultaneously at the service entrance, at the MCC bus, and at the motor terminals.
- Read the three line currents at the starter output with a true-RMS clamp meter rated for the measurement point — IEC 61010 CAT III at the MCC, CAT IV at the service entrance.
- Compute both unbalance percentages and compare every phase current against nameplate full-load current, the same comparison that settles a motor running over nameplate amps after soft starter bypass.
- Run a thermal-imaging survey at the same load, through IR windows or designated scan points, and mark any termination warmer than its peers.
- Repeat at a second load condition or time of day: unbalance that tracks other loads points to the supply; unbalance constant across loads points to fixed impedance such as a poor connection.
Findings and the decision for each case #
| Finding under normal load | What it points to | Decision |
|---|---|---|
| Voltage unbalance above 1.0% at the MCC incoming, loaded and unloaded | The service: utility supply, service transformer, service conductors or their terminations | Read at the service entrance at the same moment as the MCC; escalate both records to the utility and the site electrical engineer |
| Voltage balanced at the MCC, unbalanced at the motor terminals | Abnormal load-side impedance: termination, feeder conductor or starter pole | Infrared-scan under load, then de-energize, re-terminate and torque to the manufacturer's values |
| Voltage within 1.0% but currents spread by tens of amperes | The motor itself — winding condition or internal connection — or the meter | Winding resistance and insulation resistance tests at an outage; repeat the reading with a second calibrated meter |
| All phases within nameplate current, enclosure air hot at the relays | Enclosure cooling: blocked ventilation, clogged filters, adjacent bucket heat | Clean and restore ventilation; if controller ambient stays well above motor ambient, fit an ambient-compensated relay from the manufacturer's table — doors still closed |
| Any phase above nameplate full-load current continuously | Genuine overload: unbalance, mechanical load or machine operating state | Treat the trips as protective action; correct the unbalance and the load before touching any setting |
The record that ends the fan #
For six years the overload relays were the only devices announcing the problem, and they were cooled into silence while tower-fan motors, pumps and one hermetic compressor burned down around them. That is one story, not several: windings aged by above-limit unbalance heating with their annunciator suppressed. Proving that switchgear deterioration caused each individual failure takes the measurements above; the correction case does not wait for that attribution, because voltage unbalance above 1.0% and a 60-70 A current spread are each independently out of bounds and each requires fixing.
The maintenance side has a governing document: NFPA 70B, 2023 edition, the standard for electrical equipment maintenance. A program built on it — infrared survey of the MCC and switchgear under normal load on a set interval, torque verification and cleaning of connections at outages, protective-device testing, and written records — is what was missing, and at a government facility the survey findings plus the unbalance measurements are the documented basis for the corrective work.
Questions asked
Is it ever acceptable to run an MCC with the doors propped open and a fan on it?
No. NFPA 70, 2023 edition, Section 110.27 requires live parts to be guarded by enclosures, and a propped-open door also defeats the enclosure's ingress and fault-containment functions. The deeper problem is that the fan cools the overload relay, not the motor: it suppresses the trip while the winding keeps heating. It masks the announcement of a fault; it does not mitigate one.
Can the overload settings be raised to stop the trips?
Not on the evidence of nuisance tripping alone. First verify each heater or dial against the motor nameplate full-load current and the relay manufacturer's selection table. NFPA 70, 2023 edition, Article 430.32 caps separate overload protection at 125% of full-load current for a motor with a 1.15 service factor, and at 115% otherwise, so a correctly set relay has almost no headroom. Raising it disables the one device reporting the problem.
Why did the imbalance stay after the chiller and the MCC were replaced?
Because both replacements sit downstream of the cause. Unbalance that persists at the same points after new equipment points at the service side: the utility supply, the service transformer, service conductors, or their terminations. Measure at the service entrance and at the MCC at the same moment under the same load; the first section where the phase readings separate is the fault boundary, and everything before it is suspect.
What maintenance should this switchgear have had, and get now?
NFPA 70B, 2023 edition, is the standard for electrical equipment maintenance and the reference for the program: infrared survey of the MCC and switchgear under normal load on a set interval, torque verification and cleaning of connections during outages, protective-device testing, and written records of all of it. At a government facility those records, plus the unbalance measurements, are the documented basis for the corrective work.
Do electronic overload relays have the same ambient sensitivity?
Not to the same degree. A bimetallic or eutectic heater responds to the air around it, so enclosure temperature moves its trip point. Most electronic overload relays model heating from current alone and are ambient-compensated; confirm on the data sheet for the model in each bucket. If currents are within nameplate on every phase and the enclosure is still hot, the fixes are enclosure cooling and, where the controller ambient stays well above the motor ambient, an ambient-compensated relay selected from the manufacturer's table.