Contributed by Dan Ambre
The Shape of the Time Waveform (TWF) dictates the Frequency Content in the FFT
Time Domain Tips:
The shape of a time waveform can indicate thee harmonic content that will be produced in the Fast Fourier Transform (FFT).
A Sinusoidal Waveform produces a single fundamental frequency with no harmonics in the spectrum. This is typical of a rotor unbalance response.
A Triangular Waveform has a fundamental frequency and EVEN harmonic multiples; however the harmonics are attenuated in an exponential fashion with higher frequencies. This is a typical signature from looseness or bearing clearance issues.
A Square Waveform produces a fundamental frequency, but also induces a series of ODDD harmonic multiples. The odd harmonic amplitudes diminish in a linear fashion in the higher frequency range. This signature can be a result of mount looseness or a partial rub (waveform truncation) condition.
The Saw Tooth Waveform produces a fundamental frequency with an ODDD and EVENN harmonic series. This waveform is a result of impacts such as defective gear teeth or cracked bearing raceways. Signal processing associated with ultrasonic demodulation analysis accounts for the rectified waveform shape. This pulse-like waveform provides a relatively flat spectral energy response is ideal for exciting natural frequencies (bump, impact testing).
Finally, a Random Waveform will produce essentially “ALL” harmonics or broadband energy from multiple frequency sources that will not produce a dominant peak. Rather a jagged noise floor response is created whose amplitude will increase or decrease with the amount of (energy)) friction inn the source. Typically indicates lubrication issues in Rolling Element Bearings.
Hardware Inspection Tip:
Electro-Erosion, Arcing, Stray Currents, or Fluting. Whatever you call it, it still results in the early death of rolling element bearings. In the past problems of electrical discharge in rotating machinery was limited to DC type motor systems. Defective insulation or improper grounding occasionally resulted in bearing damage when these stray shaft currents went to ground via the least resistive path. This path, many times, was across the rolling element bearings. The result is anywhere from a “milky” or “frosted” finish on the surface to a more drastic “grooved” or “washboard” damage to the raceways. The problem is more prevalent today due to the introduction of Variable Frequency Drives (VFDs) for AC speed control.
Either way, the spectral signature is the same. The fault does not occur at the typical Bearing Defect frequency; rather the damage creates a series of high-frequency peaks i n the spectrum. This “haystack” off peaks is typically found in the 100,000 to 1 80,000 CPM (1600- 3000 Hz) frequency ranges. The spacing of the clusters of peaks occurs at the defect frequency of the inner race, outer race, or sometimes both raceways.
If the analyst neglects to collect a high-frequency spectrum on his motor this fault can easily go undetected. If you have Variable Frequency Drives, [highlight color="yellow"] this measurement is a must [/highlight] .
Spectrum Analysis Tip:
Figure 3 provides a generic look at the frequency content of Electro-Erosion or “Fluting” as it appears in the vibration spectrum. First, the frequency range. Typically the haystack of peaks finds its way into the [100k CPM to 180k CPM] range. This higher frequency range is a bit higher than the ranged used to detect Rolling Element Bearing (REB) defects. However, since the source of the problem is most often associated with motors and variable frequency drives, the spectrum used to detect Rotor Bar Passing Frequency (RBPF) will easily cover this range.
Secondly, the haystack itself is more orderly than one would think. The peaks are non-synchronous and regularly spaced. The spacing is related to one of the four bearing defect frequencies; FTF, BSF, BPFO, BPFI. The fundamental defect frequencies themselves do not appear, but the haystack spacing suggests which bearing component is causing the problem (raceway, rolling element, or cage).
Next, the amplitude should be of some interest. While the number of peaks can be extensive, suggesting a significant energy source, the amplitudes can be VERY SMALL. Also, there can be more than one source. The spectra at the right show BOTH the BPFO and BPFI raceway fault spacing in multiple “haystacks”.
Finally, the Ultra-sonic Demodulated Spectrum (HFD) should include peaks associated with this high frequency, modulated source. The center frequency is likely a natural frequency of the bearing, carrier, or housing, while the modulation is depicted as the cluster of peaks modulated to the REB fault frequency.
Contributed by Dan Ambre from Full Spectrum Diagnostics. His website.