Appendix B - Peak Detection Methods; Peak Detector Theory - GE Bently Nevada 1900/27 Operation And Maintenance Manual

Section 6 - Appendix B - Peak Detection Methods

6. Appendix B - Peak Detection Methods

Understanding the methods for peak to peak amplitude detection of a vibration
signal can help you understand why the 1900/27 Vibration Monitor is more
accurate than some other monitors and why the readings of the 1900/27 and the
Trendmaster 2000 can differ slightly under certain circumstances. This appendix
summarizes the typical methods of detecting peak to peak amplitude and then
compares the relative value of each one.

6.1 Peak Detector Theory

Three methods of peak detection are commonly used in vibration measurment
instruments - the simple peak method, the peak-to-peak divided by 2 method,
and RMS converted to peak method. The most accurate method for detecting the
peak-to-peak amplitude of vibration signals is the peak-to-peak divided by 2
method.
A simple peak detector circuit will find the maximum positive peak of a waveform.
Over some period of time, usually about one second, the peak value held by the
circuit decays so the circuit can obtain the next peak value. A peak detection
scheme using only a positive peak detector is less accurate because it assumes
that the vibration signal has equal positive and negative peaks. When measuring
an input vibration signal with a small positive peak and a large negative peak, for
example, this detector erroneously indicates a low vibration amplitude. Since
complex machine vibration signals are rarely symmetrical, a detector that
measures both positive and negative peaks is more accurate.
Peak-to-peak divided by 2 detection is a better method and is used by Bently
Nevada monitors for velocity or acceleration signals. This method finds both the
positive and negative peaks of the input signal and adds them together to get a
peak-to-peak value. Since peak value is the standard for vibration data displayed
in velocity units, the monitor divides the peak-to-peak value by two, resulting in
the peak value of the signal. The peak values held by this circuit also decay so the
circuit can detect future peaks. The peak-to-peak divided by two method ensures
that all the vibration signal from a machine, both positive and negative, is
measured and the result accurately reflects the overall vibration of the machine.
Another peak detection method is RMS converted to peak, which uses a true RMS
detection circuit to find the RMS value of the input signal, then applies a
conversion factor to find the peak value. Although some applications may call for
RMS measurements, this method is poor for determining true peak values because
the relation between RMS and peak values depends on the shape of the signal.
Typically, instruments convert RMS to peak by dividing by V2, which is only
accurate for a pure, symmetrical sinusoid. Since complex machine vibration
signals are rarely pure or symmetrical sine waves, the peak value obtained from
this method is usually in error.
43