You have a 100 HP, 460 V, 121 FLA agitator motor driving a 12:1 double-reduction gearbox with broken teeth. Vibration at blade pass, about 284 cpm, is showing up on every bearing housing on the machine. You clip a clamp-on CT on one phase, feed it through a burden resistor into the IRD/CSI collector, take a spectrum, and see nothing at 284 cpm. The question is whether current can see that load pulsation at all. It can. You are just looking in the wrong place.
Stop Looking for 284 cpm in the Raw Current Spectrum
These are the moves engineers make first on this symptom. Each one burns an hour and tells you nothing.
- Scanning 0 to 10 Hz for a 4.73 Hz peak. 284 cpm is 4.73 Hz. A torque pulsation at 4.73 Hz does not put a 4.73 Hz line in the current spectrum. It modulates the 60 Hz supply current. You will see a small hump at 4.73 Hz only if the CT or the collector input has a DC offset or the demodulation is done for you, and neither is the normal case.
- Using the vibration route settings. A route point built for gearmesh with an Fmax of several kHz and 400 or 800 lines has a bin width of several Hz. The sidebands you need are 4.73 Hz from a carrier that is 40 to 60 dB taller. They are buried in the skirt of the 60 Hz line.
- Reading amplitude in linear units. On a linear scale the 60 Hz line is the only thing on the plot. Everything you care about is in the noise floor. Switch to dB or log amplitude before you conclude there is nothing there.
- Hunting for gearmesh frequency in current. The gearbox and agitator inertia low-pass the torque ripple. Gearmesh at hundreds of Hz produces sidebands that are usually too small to separate from the floor with a clamp-on probe. That is not the fault indicator you will find first.
- Treating a blade-pass sideband as proof of broken teeth. Blade pass is a process load. A healthy gearbox on a healthy impeller shows it too. Broken teeth show up at the rotational frequency of the shaft that carries the damaged gear, not at blade pass.
Understand Why Load Shows Up as Sidebands
An induction motor draws current in proportion to slip. When load torque rises for a moment, the rotor slows a fraction, slip increases, and stator current increases. When the blade sheds its load, the reverse happens. A periodic torque ripple at frequency f_L therefore produces a periodic slip ripple and a periodic current-magnitude ripple at f_L.
Amplitude modulation of a 60 Hz carrier by f_L produces spectral lines at 60 - f_L and 60 + f_L. For your agitator:
f_L = 284 cpm / 60 = 4.73 Hz
Lower SB = 60.00 - 4.73 = 55.27 Hz
Upper SB = 60.00 + 4.73 = 64.73 HzThe 60 Hz value assumes a fixed-frequency 460 V supply, which is the normal case for this nameplate. If the motor is on a VFD, replace 60 with the actual output frequency and expect it to drift.
The same mechanism applies to every shaft in the drive train. A gear with a broken tooth loads and unloads once per revolution of its own shaft, so it produces sidebands at 60 ± k × f_shaft for that shaft and its low harmonics. That is the signature you actually want for the gearbox. Blade pass is the signature you want for the process.
Set Up the Clamp-On Current Measurement
- Clamp the CT on one motor phase at the starter or the motor junction box. One phase is enough; you are reading modulation, not balance.
- Pick a CT whose primary range covers 121 A running current without saturating. A CT sized for a small motor clips the peaks and manufactures false sidebands.
- Size the burden resistor so that the CT secondary voltage at 121 A primary stays inside the collector's rated input range:
V_burden = I_secondary × R. Read the CT ratio from the CT label and the input limit from the collector manual. Do not guess either. - Check the CT's low-frequency response on its datasheet. Some clamp-on probes roll off below a few Hz, which does not matter here because the information is at 55 and 65 Hz, but a probe that is noisy near 60 Hz will mask the sidebands.
- Take a matching voltage measurement on the same phase if the collector has a second channel. A supply voltage that is itself modulated at 4.73 Hz produces the same sidebands with a perfectly healthy load. Voltage sidebands present means the source is the supply, not the agitator.
Configure the Collector for Sideband Resolution
Bin width sets whether you can see this at all. Use the relationship:
bin width = Fmax / lines
sample time = lines / Fmax (= 1 / bin width)- Set Fmax between 100 Hz and 200 Hz. Nothing above 65 Hz matters for the blade-pass sidebands.
- Use a Hanning window and at least 4 averages with no overlap. Load sidebands are small and random process noise is not; averaging is what pulls them out of the floor.
- Display amplitude in dB referenced to the 60 Hz line. Read sideband height as a negative dB number relative to the carrier.
- Run the agitator at its normal fill level and viscosity. A half-empty vessel changes the torque ripple and the sideband amplitude, and the reading is not comparable to anything.
The resolution above is finer than the 4.73 Hz spacing strictly needs. You want it anyway so that output-shaft, intermediate-shaft, and pole-pass sidebands land in separate bins from the blade-pass pair.
Read the Spectrum: Blade Pass Versus Broken Teeth
Before you interpret anything, pin down the shaft speeds. Read motor RPM from the nameplate or a strobe under load. Output shaft frequency is motor speed divided by 12. The intermediate shaft depends on how the 12:1 is split between the two stages; read it from the gearbox drawing. Then count the impeller blades. Blade count × output shaft speed must equal 284 cpm. If it does not, either the blade count or the assumed motor speed is wrong. As an example only: a 4-pole motor near 1780 rpm gives an output near 148 rpm, and 284 cpm is close to twice that, pointing at a two-blade impeller; a 6-pole motor near 1140 rpm gives about 95 rpm, and 284 cpm is close to three times that. Measure, do not assume.
| Sideband pattern around 60 Hz | Most likely cause | Confirmation |
|---|---|---|
| ± 4.73 Hz (blade pass), present on all agitators | Normal process load pulsation | Same pair on a known-good unit; tracks fill level |
| ± output shaft frequency and harmonics, one unit only | Damaged gear on the output shaft | Vibration 1× output shaft and gearmesh sidebands at the same spacing |
| ± intermediate shaft frequency and harmonics | Damaged gear on the intermediate shaft | Vibration 1× intermediate shaft; borescope the second stage |
| ± motor shaft frequency | Coupling, misalignment, or motor-side pinion | Vibration 1× and 2× motor at the coupling end |
| ± twice slip frequency, close in to the carrier | Rotor bar or end ring, not the gearbox | Sideband height in dB versus carrier; separate MCSA rotor test |
| Identical sidebands also in the voltage spectrum | Supply modulation, not the load | Check upstream loads on the same bus |
Do not read a single sideband height as a go/no-go value. Load sidebands scale with inertia, damping, and process conditions that differ between agitators. Compare each unit to the others and to its own history.
Verify the Reading and Trend It
- Take the same current spectrum on an agitator with a known-good gearbox at the same process condition. The blade-pass pair should be present on both. Shaft-frequency sidebands should be present only on the damaged unit.
- Overlay the current spectrum with the vibration spectrum from the gearbox housing. Every sideband spacing you claim in current must match a 1× shaft line or a gearmesh sideband spacing in vibration. If it does not, you are looking at noise.
- Change the load. Raise or lower fill level or viscosity and re-take. The blade-pass sidebands should move in amplitude with load. A pair that does not respond to load is not load.
- After the gearbox is repaired, re-take with identical settings. The shaft-frequency sidebands must drop into the floor. The blade-pass pair stays; it is your new baseline.
- Save the setup as a dedicated route point with the Fmax, lines, window, averages, CT ratio, and burden value recorded. Trending is worthless if the next reading uses different resolution.
FAQ
What happens if I set Fmax to 1 kHz like a vibration point?
The 55.27 and 64.73 Hz sidebands sit inside the leakage skirt of the 60 Hz line and you will report a healthy motor regardless of what the gearbox is doing.
What happens if the clamp-on CT saturates at 121 A?
Clipped peaks generate odd harmonics and intermodulation products across the whole spectrum, including false lines near 55 and 65 Hz. Check the CT primary rating before you trust any sideband, and confirm the burden voltage is inside the collector input limit.
What happens if I see the blade-pass sidebands but no shaft-frequency sidebands on the damaged unit?
The broken-tooth torque pulse is being absorbed by drive-train inertia and is below the floor of a single-phase clamp-on measurement. Increase averages, tighten the bin width, and rely on the gearbox housing vibration for the tooth fault; keep the current sidebands for load trending.
What happens if the same sidebands appear in the voltage spectrum?
The modulation is arriving from the supply, usually from another cyclic load on the same bus, and the current sidebands say nothing about your agitator. Find the upstream load first, or measure at a time it is off.
What happens if the motor is on a VFD?
The carrier is the drive output frequency, not 60 Hz, and it moves with the speed reference; recompute the sideband positions from the actual output frequency and use a CT and burden rated for PWM current. If the sidebands still cannot be separated from the drive's own harmonics, stop and take the case to the collector manufacturer's applications support with the raw time waveform, the CT model and ratio, the burden value, and the nameplate data. If the gearbox is under warranty, send the correlated vibration and current spectra to the gearbox manufacturer's service group before opening it.