Figure 8: Setup for testing the ACD.
The acoustic data was recorded on the computer as a time series. To record time series data effectively, the frequency band of interest (fn), a desired error level (sr), and a desired frequency resolution (Df) must be defined. Through operational experience (and to make a long story short), these parameters were set at 300Hz, ten percent, and 0.5Hz respectively. Since aliasing is to be avoided, the sampling frequency (fsample) must follow the Nyquist criterion. Therefore, the minimum sampling frequency should be at least 750Hz. However in reality, a sampling frequency of 3 kHz was used to capture not only the 300Hz signal but also any harmonics of the system that might exist. The sampling duration (T), which is inversely related to Df (T = 1/Df), was set at 2 seconds to obtain the desired frequency resolution. It should also be noted that the buffer size of the computer A/D board also must be set. This is based on the number of points obtained in the waveform and the time period over which the waveform is taken. The buffer size chosen in this case was 663kpts. A series of two data sets of ambient noise in the tank and a series of three data sets of noise with the acoustic conditioning device active were acquired.