Are there any significant differences between the well known
methods of resonance demodulation and envelope detection that are widely used in
rolling element bearing diagnostics?
(The question was asked by the participants of "Vibration diagnostics of
rotating machine and equipment" seminar)
The question is answered by
Alexej Barkov:
These two methods have practically the same technical and
hardware realization but have very different physical basics in diagnostics. The
method of resonance demodulation investigates the natural oscillations of the
bearing elements (fixed race, rolling elements and other) that are excited by
separate shock pulses which appear as a result of faulted rolling surfaces
contacts. The algorithm of investigation includes the following operations:
natural bearing element oscillation frequency
search;
extraction of the natural oscillation from the vibration
signal, for example by a frequency band filter;
enveloping of the extracted part of the signal, for
example by detection with the sequential filtration of the low frequency
part of the detected signal;
spectral analysis of the envelope;
the analysis of the frequencies and amplitudes in the
spectra of the envelope.
The envelope method used in our systems investigates the
oscillations excited by both the friction forces and shock pulses. The
investigation algorithm differs practically only in the first and the last
operations and includes:
search of a signal frequency band where there are no
resonances and harmonic components (to tune away from natural and harmonic
oscillations);
extracting high frequency random vibration components with
a band pass filter;
enveloping of the extracted part of the signal, for
example by detection with the sequential filtration of the low frequency
part of the detected signal;
spectral analysis of the envelope;
calculation of the modulation amplitude of the harmonic
components of the envelope spectra and analysis of the frequencies and
amplitudes of the random vibration modulation.
The difference in these two methods is in that in the first case the natural
oscillations are investigated, whereas in the second case the forced
oscillations of the bearing elements are investigated. The natural oscillations
give maximum information about the features of the oscillating system, for
example, about the rigidity, mass, losses in the oscillation system, and only
limited information about the oscillating forces that define the initial
deflection of the oscillation system from the equilibrium point whereas the
forced oscillations give maximum information about the oscillating forces but
minimum information about the oscillating system. The rolling element bearing
vibration diagnostics is, first of all, the analysis of the features of the
oscillating forces but not of the oscillation system that has many individual
peculiarities for each bearing unit in each certain machine. From the above, it
follows that the envelope method has certain advantages for the rolling element
bearing diagnostics but often investigators try to use the resonance
demodulation method instead of the envelope method. The question is why? There
are technical and physical reasons. The technical reason lies in the fact that
the resonance increases the contribution of the investigated components in the
measured signal by dozens of times and allows extraction of them from the total
signal without alteration, even with instruments made many years ago that have
very narrow dynamic range and bad linearity. Modern digital analyzing
instruments have both the dynamic range and the linearity sufficient to avoid
the use the mechanical resonance to gain the signal and not loose the major part
of the diagnostic information. The physical reason lies in the fact that, if
there are harmonic components in the part of the signal that is extracted by the
filter for further enveloping, then it corrupts the diagnostic information
received from the envelope spectrum. The resonance of the bearing elements gain
the random vibration components and thus decrease the negative influence of the
harmonic components that coincide with the filter frequency band in the features
of the measured envelope spectrum. But, if we consider this fact and take into
account the capabilities of modern instruments for signal analysis, then we must
say that the harmonic components can be extracted from the signal before
enveloping using for example adaptive filters. So, modern measuring and signal
analyzing techniques enable using completely the advantages of the envelope
method to analyze the constrain forces in the friction units not only of
mechanical nature but also of air and hydrodynamic origin. This enables solving
the friction units diagnostics in more detail than the methods of investigation
the natural oscillations excited by the shock pulses. And the last
consideration. The method of resonance demodulation is patented (the USA patent
of 1974). The envelope method is patented in the USSR in 1978. At least the
second patent is already expired. But one should take into account that in the
USA there is a patent # 5477730 issued 12.26.95 that concerns the method of
separating the harmonic and random vibration components before enveloping that
allows the use of the envelope method without the need for concern for the
presence of harmonic components in the measured data. We have employed this last
technique in data acquisition drivers for hardware from VSC and Data Physics for
our DOS DREAM and have similar drivers planned for our DREAM for Windows
software for the same manufacturer's current hardware. More information on this
technique can be found in our on-line library of technical papers in the
Bearing, Data Acquisition section.
