MCSA

MCSA Fault Severity and Alarm Thresholds

How to interpret sideband amplitude in MCSA spectra, what dB levels suggest developing faults, and why trending over time is more reliable than applying fixed thresholds.

Key Principle

A sideband at -48 dB on a motor where the baseline was -65 dB has risen 17 dB, that is a significant change. The same -48 dB on a motor where the baseline was -45 dB may be normal. Alarm thresholds in MCSA must be relative to a known baseline, not absolute values applied universally.

Sideband Amplitude (dB below fundamental) Healthy Below -60 dB — routine monitoring Minor -60 to -54 dB — establish trend, increase interval Early Fault -54 to -48 dB — monthly monitoring, plan inspection Developing -48 to -40 dB — weekly monitoring, schedule outage Severe Above -40 dB — engineering assessment, consider load reduction baseline -60 dB -54 dB -48 dB -40 dB above MCSA sideband severity thresholds for rotor bar faults at full load. Exact thresholds vary by motor type, baseline condition, and operating load.

What the dB Scale Means in MCSA

In motor current signature analysis, amplitude is expressed in decibels relative to the supply frequency component. The fundamental supply frequency (50 Hz or 60 Hz) appears as the dominant peak in the current spectrum. All other frequencies, including fault sidebands, are measured relative to that peak.

A reading of -54 dB does not mean the sideband is at an absolute power level of -54 dBm or any other absolute unit. It means the sideband is 54 dB below the magnitude of the supply frequency component in that same spectrum. This is a critical distinction: the same motor under different load conditions will produce different supply frequency amplitudes, and therefore different absolute sideband levels, even if the fault condition is unchanged.

The practical consequence is that you cannot directly compare a reading taken at 40% load to one taken at 85% load. The dB reference changes with load, so the sideband level changes even when the fault does not.

Commonly Cited Absolute Thresholds and Their Limitations

Several alarm levels appear repeatedly in MCSA literature and diagnostic tool documentation:

These values originated primarily from research on large induction motors under controlled conditions. They are useful as orientation points, not as universal standards. Two problems arise when applying them directly:

First, the thresholds assume full-load operation. Sidebands at partial load may fall below the threshold not because the motor is healthy but because lower slip reduces the sideband amplitude. A -58 dB reading at 50% load may represent the same physical damage as -46 dB at full load.

Second, motor design affects baseline amplitude. Older motors, motors with manufacturing variation, or motors driving loads with inherent torque pulsation (compressors, reciprocating pumps) may show sidebands well above -54 dB with no fault present. Applying the -54 dB threshold to these machines generates nuisance alarms and erodes trust in the monitoring program.

The Baseline-Relative Approach

A more reliable method is to establish a healthy baseline for each motor individually and alarm on deviation from that baseline rather than on absolute amplitude.

The process has three steps:

  1. Measure the motor under known, stable conditions shortly after installation or a confirmed healthy inspection. Record the sideband amplitudes for all frequencies of interest, and record the operating load at the time of measurement.
  2. All subsequent measurements must be taken at the same load conditions, within a reasonable tolerance (typically ±5% of rated load for rotor fault analysis).
  3. Calculate the change in sideband amplitude compared to baseline. A 3 dB rise requires awareness. A 6 dB rise warrants investigation. A 10 dB or greater rise over a short period signals a rapidly progressing fault.

The absolute level still matters. A baseline at -30 dB is already abnormal and means the motor either has a pre-existing fault or the measurement conditions are not suitable for MCSA analysis. The baseline-relative approach works on top of, not instead of, a sanity check on the initial reading.

Measurement Repeatability Requirements

Baseline-relative analysis depends entirely on measurement consistency. The following conditions must match across measurements for valid comparison:

Load Dependency of Sideband Amplitude

Rotor bar fault sidebands appear at frequencies symmetrically around the supply frequency: at (1 +/- 2ks) x f1, where k is an integer, s is slip, and f1 is supply frequency. Slip increases with load. As slip increases, the sideband frequency moves further from the supply frequency and the amplitude of the fault sideband rises.

This means a rotor bar fault may be invisible at very light load. The motor draws low current, slip is minimal, and the sideband sits close to the supply frequency peak where it is masked by spectral leakage. The same fault under full load produces a clear, measurable sideband at a distinct frequency.

The practical rule: always record the actual measured load current alongside any MCSA result. Without load information, an individual measurement has limited diagnostic value. Two measurements from the same motor at different loads cannot be compared without load normalization.

Severity Tables by Fault Type

Rotor Bar Sidebands (at Full Load)

Sideband Amplitude (dB below supply frequency) Condition Indication Recommended Action
Below -60 dB Healthy rotor; no indication of broken bars Continue routine monitoring interval
-60 dB to -54 dB Minor asymmetry; may be normal for some motor designs Establish trend; increase frequency to quarterly
-54 dB to -48 dB Early fault indication; one or two bars potentially cracked Monthly monitoring; plan inspection at next outage
-48 dB to -40 dB Developing fault; multiple bars affected Weekly monitoring; schedule outage within 60 days
Above -40 dB Significant rotor damage Engineering assessment; consider load reduction pending repair

Eccentricity Sidebands

Static and dynamic eccentricity produce sidebands at (1 +/- R/p) x f1, where R is rotor slot number and p is pole pairs, and at other combinations depending on motor geometry. These sidebands are present to some degree in all induction motors. Typical healthy amplitude: -50 dB to -70 dB depending on motor design and pole count.

Sideband Rise from Baseline Condition Indication Recommended Action
Less than 4 dB Normal variation; no change in eccentricity Continue current interval
4 dB to 8 dB Possible bearing wear or shaft deflection Cross-check with vibration; increase measurement frequency
8 dB to 14 dB Progressive eccentricity; likely bearing or alignment issue Bearing inspection within 30 days; check alignment
Above 14 dB Severe eccentricity; risk of stator-rotor contact Engineering review before next restart

Bearing Fault Frequencies in Current

Mechanical bearing faults (BPFO, BPFI, BSF, FTF) modulate motor torque and therefore appear in the current spectrum at the bearing fault frequency and its sidebands around the supply frequency. Current-based bearing detection is less sensitive than vibration-based detection. Bearing faults typically only appear clearly in current spectra when the fault is moderate to advanced.

Current Sideband Amplitude at Bearing Fault Frequency Condition Indication Recommended Action
Not detectable in current spectrum Early-stage bearing fault or healthy bearing Rely on vibration for early bearing detection
-60 dB to -50 dB above baseline Developing bearing fault visible in current Confirm with vibration analysis; plan bearing replacement
Above -50 dB or rapid rise Advanced bearing fault Expedite bearing replacement; risk of secondary damage

Comparison: Baseline-Relative vs Absolute Alarm Methods

Factor Absolute Threshold Baseline-Relative Threshold
Setup requirement None; apply published values directly Requires documented baseline at known load
False alarm rate on normal motors with high inherent asymmetry High Low; baseline absorbs inherent asymmetry
Sensitivity to load variation High; load changes produce alarm threshold crossings unrelated to fault Controlled by requiring consistent measurement load
Ability to detect gradual progression Detects only after crossing fixed threshold Detects trend before threshold; earlier warning
Useful for single-measurement assessments Yes; provides immediate rough classification No; requires at least two data points for trend
Best application Initial screening; motors with no history Ongoing condition monitoring programs

Practical Response Triggers

Severity tables define what the data says. Response triggers define what you do about it. These two things are separate decisions, and the response should account for motor criticality, maintenance schedule, and operational constraints.

First Alarm: Increase Measurement Frequency

When a sideband crosses an early warning threshold (or rises 6 dB above baseline), increase the measurement interval from quarterly to monthly. Do not reduce load or schedule downtime yet. The goal is to determine whether the condition is stable or progressing. A stable reading over several months at an elevated level may simply be a new baseline.

Second Alarm or Rapid Progression: Schedule Inspection

If the sideband continues rising across multiple monthly measurements, or if it jumps more than 6 dB between any two consecutive readings, schedule a physical inspection at the next available planned outage. A rapid rise (more than 10 dB over 30 days) indicates an actively progressing fault and should be treated urgently regardless of current absolute level.

Critical Level: Engineering Assessment Before Restart

When sidebands reach the highest severity tier or when multiple fault signatures appear simultaneously, the motor should not be restarted after the next shutdown without an engineering review. This does not necessarily mean immediate shutdown if the motor is running and the load can be safely continued, but it does mean the motor should not be cold-started into a high-load application until the fault is assessed.

For motors driving safety-critical equipment or processes where failure causes production loss exceeding the cost of a planned repair, the threshold for engineering assessment should be set more conservatively than the generic tables above suggest.

Continuous MCSA Monitoring with Tractian

Tractian's condition monitoring sensors capture motor current signatures continuously, trend sideband amplitudes automatically, and alert your team when fault levels change, not just when they cross a fixed threshold.

See Electrical Monitoring
The thresholds and guidance on this page are reference values for educational purposes. Alarm levels for specific motors should be established in consultation with the motor manufacturer, a qualified electrical engineer, and site-specific operating conditions. Do not rely on generic thresholds as the sole basis for operational decisions on critical equipment.