A 1250 kW HT machine back from rewind, uncoupled, drawing 40 A, sitting at 60 C with an audible hum. Two symptoms, and they usually share one root. The instinct is to start adjusting. Look at the trend first, and confirm what the 60 C number is attached to, because half the calls that start this way end with a measurement error and the other half end with a coil group that came back from the winding shop connected the wrong way round.
Which fixes get tried first, and why they fail
Re-greasing the bearings is the reflex when a large machine runs warm. It fails because a bearing that is genuinely running hot shows a rising trend at the bearing RTD or the housing, not a stator reading, and no-load bearing losses on a machine this size do not lift a winding 60 C. Grease before you know which sensor moved and you have changed the machine without learning anything.
Swapping two supply leads to reverse rotation, or re-torquing the terminal box lugs, addresses a real class of fault but not this one. Loose HT connections show up as flashover, unbalanced current or a heated lug, not as a symmetric hum. Torque them anyway during inspection, then move on.
Dynamic balancing the rotor gets proposed the moment somebody calls the noise "vibration." A hum at twice supply frequency is electromagnetic and does not respond to balance weights. Balancing a rotor to cure a magnetic hum wastes a shift and leaves the fault in place.
Extending or shortening the no-load run to "see if it settles" is the last of the wrong moves. Without a logged trend you cannot tell a starting-current transient from a genuine steady rise. On an uncoupled start the winding temperature can peak shortly after the inrush and then fall back; read one spot value at the wrong minute and you draw the opposite conclusion.
What is 60 C actually measuring?
Absolute temperature tells you nothing on its own. Rise above ambient is the quantity that matters. At 40 C ambient in a plant hall, 60 C is a 20 K rise on an unloaded machine, which is unremarkable. At 20 C ambient it is a 40 K rise with no load torque and no rotor current worth speaking of, and that demands an explanation.
Then confirm the sensor. Establish whether the 60 C came from an embedded stator winding RTD, a bearing RTD, a frame-mounted thermocouple, or a handheld pyrometer aimed at the frame. Check that the monitoring device is configured for the RTD type actually installed, that the three-wire lead compensation is intact, and that the field terminations are tight. A high-resistance joint anywhere in an RTD loop adds apparent temperature; a 100-ohm-class sensor read as if it were a copper element reads high across the whole range. Verify the reading against a second instrument on the same body of metal before you accept it.
Signals to measure before you touch anything
| Signal | Source | Symptom when the value is wrong |
|---|---|---|
| Stator winding temperature | Embedded RTD, at the monitor and at the marshalling box | Loose lead or wrong sensor type configured reads high with no real heat in the copper |
| Bearing temperature | Bearing RTD or housing | Rising trend points at lubrication, alignment or preload, not at the stator |
| Line-to-line voltage, all three pairs | At the motor terminals, not at the switchgear | Unbalance drives negative-sequence current, heating and a hum with no load applied |
| Phase current, each phase separately | Clamp or CT per phase during the no-load run | Unequal phase currents indicate a winding connection error, shorted turns or supply unbalance |
| Winding resistance, phase to phase | Micro-ohmmeter, machine cold and isolated | One phase out of step with the other two exposes a reversed group, wrong tap or a shorted section |
| Air gap, four points | Feeler gauge, top, bottom and both sides | Eccentric gap produces unbalanced magnetic pull, a beat in the noise and localised heating |
| Vibration spectrum | Accelerometer on the bearing housings | A peak at twice supply frequency is magnetic; 1x running speed is mechanical |
Take the electrical readings during the same run that produced the 60 C, not on a separate trial. The 40 A figure is meaningful only against the machine rating: at 6.6 kV, a 1250 kW motor draws roughly 135 A at full load assuming 0.85 power factor and 0.95 efficiency, so 40 A is about 30 percent of rated current and sits in the normal no-load band. At 11 kV the same machine draws about 81 A at full load, and 40 A is close to half of rated current with nothing on the shaft, which is high enough to be the fault itself. Read the nameplate voltage before interpreting the 40 A.
Why does a rewound stator hum at no load?
With the shaft uncoupled there is no rotor current of consequence and no load loss. Whatever heats the machine has to be magnetising current, iron loss, or circulating current inside the winding. The hum comes from the same place: forces in the magnetic circuit at twice supply frequency, or a rotating force from an asymmetric field.
The dominant candidate after a rewind is a reversed coil group. Connect one group backwards and its MMF opposes the rest of its phase. The phase currents go unequal, the airgap flux distribution loses symmetry, magnetising current rises, and the machine hums under a force wave that the design never intended. The copper heats even though there is no shaft load, and the noise is present from the instant the contactor closes.
Shorted laminations are the second candidate, and on a rewound core they are common: the old winding was burned or cut out, the slot walls were damaged, and the interlaminar insulation broke down. The short circuits axial flux paths, eddy current flows in the shorted plates and produces a local hot spot with no load current anywhere near the machine. That is the classic "hot with nothing connected to the shaft" fingerprint.
Behind those two sit a loose stator or rotor core, an unbalanced air gap from bearing housing or endshield misalignment, an incorrect star/delta or series/parallel connection in the terminal box, supply voltage unbalance, and broken rotor bars or end ring joints. With six leads brought out, one crossed pair between line and star point end produces exactly this picture. Any of these can occur together, and often more than one is present after a rewind.
The sequence that isolates it
- Log the no-load run instead of spot reading it. Record winding RTDs, bearing RTDs, ambient, and the three phase currents at fixed intervals. Decide whether the temperature is climbing, flat, or falling back from an inrush peak.
- Measure all three line-to-line voltages at the motor terminals during the run. Calculate the unbalance. Supply unbalance produces negative-sequence current and heating that no work on the machine will fix.
- Clamp each phase separately. Equal currents with a hum push you toward the core and air gap. Unequal currents push you toward the winding connection.
- Isolate, prove dead, and measure winding resistance phase to phase with a micro-ohmmeter. Compare the three values against each other and against the rewinder's test sheet. A group reversal or an internal short shows here.
- Run insulation resistance and polarisation index, then a surge comparison test on each phase. Surge testing is what finds a reversed group or turn-to-turn short that resistance measurement is too coarse to see.
- Check the air gap with feeler gauges at top, bottom and both sides, at both ends. Compare the four readings; an eccentric gap explains both the pull and the beat in the noise.
- If the winding tests clean, energise the core with a flux ring loop test and scan the bore with a thermal camera. Hot spots on the tooth tips or in the back iron confirm shorted laminations.
- Confirm the terminal box connection against the nameplate and the rewinder's connection diagram, lead by lead, before reassembly.
How do you verify the repair held?
Whatever the finding — group re-terminated, laminations separated and treated, core re-tightened, endshield re-shimmed for concentric gap — repeat the no-load run under the same instrumentation and hold it long enough to reach thermal equilibrium, not just long enough to feel warm. Record ambient at the same time and report rise, not absolute.
Three results have to line up. Phase currents balanced within a few percent of each other. Winding rise stable and comparable to the rewinder's own no-load test if they ran one — ask for that test sheet, including run duration and whether a steady trend was reached. Hum reduced to the normal magnetic note of a machine of this frame, with no twice-line-frequency peak dominating the vibration spectrum. If any one of those three is still out, the fault was not the one you repaired.
Tuning does not fix wiring, and a satisfactory no-load run does not license a full-load start on its own. Bring the machine up on load in steps with the RTD trend live in front of you.
Stop and escalate when the surge test fails a phase, when the core loop test shows a hot spot, or when the winding resistances do not match: those are workmanship findings that belong back with the rewind shop under the repair warranty, not on the shop floor with a megger. Involve the motor OEM when the air gap is eccentric beyond the drawing tolerance or when the core has to be restacked, because that touches the magnetic design. Do not couple the load to prove a point while any of those is open.
Frequently Asked Questions
How do I tell whether a humming motor has an electrical or a mechanical fault?
Record a vibration spectrum on the bearing housings and note what happens at the instant supply is removed. A magnetic fault peaks at twice supply frequency and the noise disappears the moment the contactor opens; a mechanical fault sits at 1x running speed and coasts down with the rotor.
How do I confirm a reversed coil group after a rewind?
Measure phase-to-phase winding resistance with a micro-ohmmeter and compare the three values, then run a surge comparison test on each phase. A reversed group shows as an outlier resistance and as a diverging surge waveform, and it usually comes with unequal no-load phase currents.
How do I check for shorted laminations on a large stator core?
Energise the core with a flux ring loop test at rated flux density and scan the bore with a thermal imager, or use a low-flux core test instrument. Hot spots on tooth tips or in the back iron identify the shorted plates and their axial position.
Is 40 A no-load current normal on a 1250 kW HT motor?
It depends on the nameplate voltage. At 6.6 kV, full-load current is roughly 135 A assuming 0.85 power factor and 0.95 efficiency, so 40 A is about 30 percent of rated and typical; at 11 kV, full-load current is around 81 A and 40 A no-load is high enough to indicate a magnetic circuit or connection problem.