Sonar Index

Simulation of Sidescan Transducer Arrays

Page: 1,  23

A sidescan sonar ranks among the most desired tools of those dealing with the investigation of the seafloor, for instance for diveground-prospection or shipwreck-hunting, but unless you do this professionally the prices of this kind of equipment are out of reach. Without doubt a sidescan sonar is a precision instrument, but on the other hand the basics are long known. So there are a handful of amateurs around the world that build their own sidescan sonars, some do it in a really professional way, and some go the easier but nevertheless promising way of using an array of fishfinder or depthsounder transducers to build a sidescan sonar transducer.
Nowadays you can get very nice readout electronics for fishfinders at reasonable prices, so the most important question is what can be expected from such transducer arrays compared to professional sidescan transducers. I ran some simulations to find an answer and to see if the fishfinder transducer arrays can be optimized.

The first simulation (Fig.1b) shows what we would like to have: the theoretical sound-pattern of a nearly perfect sidescan transducer array working at 50kHz. The calculated array is 610mm long, consisting of 6 inividual piezo-elements measuring 100mm x 25mm at the side facing the water, the elements are spaced at 2mm (Fig. 1a).

sidescan sonar transducer
Figure 1a: Sidescan Sonar Transducer Array, 610mm x 25mm

Sidescan Transducer: Polar Plot
Figure 1b: polar plot for the calculated horizontal energy distribution of the sidescan transducer in Fig.1a working at 50kHz.
89% of the emitted energy is concentrated in the central lobe.

The horizontal beamwidth is determined by the length of the transducer (610mm), the vertical beamwidth (not visible in the plot of Fig.1b) by the transducer hight (25mm).
Another kind of plot is also very instructive: Fig.1c shows the two-dimensional soundfield of the transducer that is projected into the water as a grayscale image as seen from the position of the transducer.
2d soundfield of a sidescan transducer

Figure 1c: 2-D projected soundfield of the transducer in Fig.1a Fieldsize: 120° horizontally (azimuth) x 120° vertically (hight)

The second simulation (Figures 2b and 2c) shows what we have when we use a fishfinder transducer with a single 44mm circular piezo crystal (Fig. 2a). A very broad circular beampattern is sent away from the transducer.
single fishfinder transducer
Figure 2a: geometry of the sound projecting plane of a single fishfinder transducer

polar plot of single fishfinder transducer
Figure 2b: polar plot for a single fishfinder transducer (Fig.2a) working at 50kHz.

2d soundfield of a fishfinder transducer

Figure 1c: 2-D projected soundfield of a single 44mm fishfinder transducer at 50kHz. Fieldsize: 120° x 120°

The good resolution of a sidescan sonar among other things is due to a small central lobe, so we want a long transducer. It is clear that a single fishfinder transducer will never draw a detailed reproduction of the seafloor. The idea when working with fishfinder or depthsounder transducers is to set them up in a row to imitate a long transducer. How will the beampattern of such a setup look like and can it be optimised?
Next Page