Acoustic measurement allows engineers to assess the sound related to a device under test, predict acoustic environments, and address design problems. However, this measurement does not contain frequency information, making it unsuitable for comparing sound and vibration. To address such, vibration test engineers use a frequency analysis technique called octave analysis.
What is Octave Analysis?
Octave analysis allows for the objective evaluation of sound and vibration by grouping the frequencies of an acoustic signal into “bins,” often to reflect how humans perceive the frequency spectrum.
First, a software program filters the acoustic signal and measures the sound pressure levels at the output. The engineer can then apply averaging and weighting techniques in the frequency domain to correspond to the measurements of their desired evaluation of sound.
The standard audible frequency range for humans is 20-20,000Hz, which can be filtered for analysis. When an acoustic signal’s frequency range is fractioned into eight bins, engineers refer to the sections as octave bands. For an arrangement of twenty-four frequency bins, they are called one-third octave bands. A fractional-octave analysis allows the engineer to select a frequency resolution that is well-suited to their signal of interest.
As the human ear responds to frequency measurement on a logarithmic scale, the frequency content is measured logarithmically, and sound levels are in decibels (dB). Octave analysis requires three components: sensors, a data acquisition system, and software analysis.
Constant Percentage Bandwidth (CPB)
For filtering the frequency spectrum, the software computes the spectral amplitude of the logarithmic frequency bins in proportion to the center frequency of that bin. The amplitude of each bin represents the intensity of the signal in that frequency range and is the sum of the range’s RMS values.
These filters are often called constant percentage bandwidth (CPB) filters because the filter’s bandwidth is a fixed percentage of the band’s center frequency. The most common CPB filters are those with a one-third octave bandwidth, but more frequency bins provide a more detailed analysis of noise content.
Filter-Based vs. FFT Octave
As discussed, octave analysis applies bandpass filters to the frequency range and averages each filter’s output to compute its power. The values are then displayed on a bar graph. The industry often refers to this process as true octave analysis, and it produces more accurate results.
However, some software programs can also use FFT data—which measures the frequency content linearly—and assign the energy to the proportional octave. This method is an efficient option when the engineer wants the spectrum values without the complexity of filter-based analysis. Still, FFT is only recommended if your computer cannot handle the filtered option.
Frequency Weighting and Averaging
The human ear does not perceive changes in sound pressure as loudness because of low sensitivity at low and high frequencies. Engineers apply frequency weighting to filtered acoustic signals to more closely represent the human response to acoustics.
The International Electrotechnical Commission (IEC) developed a standard set of frequency-weighting curves, including A, B, C, and D weight. Many standards include frequency weighting for occupational and environmental noise. The weighting curve selection depends on the type of measurement. For example, the A-weighted curve is ideal when humans are involved with the acoustic signal.
An engineer may also apply averaging to a filtered signal for a more stable representation of the true signal.
Octave Analysis in ObserVIEW
ObserVIEW 2021.2 generates octave bands with an 8th-order filter to meet IEC 61260-1 Class 1 filter specifications. Fast/slow or user-defined time weightings, linear or exponential averaging, and peak hold options are available. A, C, and Z frequency weighting options are also available to meet IEC 61672-1 requirement.
The software includes the most used fractional octave bands; however, the user can enter any 1/N fraction that suits their test objectives. ObserVIEW does not have a limit on fractional octave bands. (Note: there is a soft limit at 1/96 for computer performance, but the user can override the limit.)
The octave band graph supports the advanced functionalities of ObserVIEW, including live analysis, copy-and-paste, and graph traces.
Vibration Research’s data acquisition systems offer the functionality to acquire and analyze acoustic signals. All VR hardware includes a BNC input that supports a microphone and is capable of data acquisition. If your testing lab is already using VR hardware for vibration testing, sound acquisition and analysis is an economical addition.
The root mean square (RMS) is a measurement of intensity (amplitude). An octave band plot splits a signal into bins with octave-spaced center frequencies. The amplitude of each bin represents the signal intensity in that frequency range. The amplitude is calculated as the sum of the range’s RMS values.
The Overall RMS displays the total RMS for the integration period. For pressure units, the measurement is Overall SPL (sound pressure level), which provides the total volume of an audio signal.
Selection Range Average
The selection range average generates the octave band plot using the average amplitude values over a selected data range.
The integration time is the time constant in the exponential moving average, which the software uses to generate the octave bins. For playback, a longer integration time results in slower changes in the plot but less pronounced peaks. In post-processing, the integration time calculates the linear RMS values.
A larger integration time will use a wider time data range to calculate RMS values. The suggested integration times (Fast – 0.125 sec and Slow – 1 sec) are preset values that technicians can use to compare the effects of different integration times. The integration time can also be set manually.
Octave Analysis Software Page