The most crucial aspect of this system to investigate in order to determine the feasibility of using sonic radar is the reception of a pulse that has been reflected off a surface at least 40m away. In order to determine whether this solution was realizable, an experiment was setup to mimic this aspect of the solution.
The first design decision of the experiment was to choose frequency and pulse width for the signal. It was decided that a 10 kHz frequency, with a pulse envelope width of 0.5ms, would be sufficient. This frequency is low enough that it wonıt be attenuated before it can be received at distances of 40m, but high enough that the measurement will still be reasonably accurate. Lower frequencies result in less accuracy because the wavelength is very large when compared with the distance to be measured.
The experiment involved using a function generator to create the pulse. This was then passed through a power amplifier, and then fed into a high frequency, audio quality speaker (tweeter). Mounted above the tweeter was a directional microphone to receive the echo. The output of the microphone was connected to a bandpass filter with a large gain factor. The final output of the filter was viewed on an oscilloscope to determine if an echo had been received. The experiment setup is illustrated in the diagram below.
Figure 1: Experiment Setup Overview
To ensure that the signal is not totally attenuated before the echo reaches the microphone, the original pulse emitted from the speaker must be sufficiently loud. An equation describing total attenuation can be seen below:
Total Attenuation = 20log2H + a2H
Where a is the absorption in dB/m and H is the height.
In the experiment, a signal strength of 10V peak to peak was generated with the function generator. This value was used because it is the largest signal obtainable from the Agilent Technologies function generators available in the lab. A power amplifier was used to boost the current to the speaker, and this system delivered a total power of approximately 12.2W to the speaker.
The power used to drive the speaker could have been increased by using a more suitable power amplifier. With an input of 10V peak to peak, the power amplifier was clipping the signal
The chosen speaker must be designed for use in the high audio range. It should also be fairly directional, so that echoes are only received from objects that are in front of the speaker.
The speaker used in the experiment was a Philips AD0163. This is a tweeter, designed for use in a home stereo system. It has a maximum power handling capability of 15W, with a sensitivity rating of 92 dB with 1W input. When used in the experiment, powered with 12.2W, the resulting output from the speaker was calculated at approximately 103 decibels. Unfortunately, this speaker has a beam width of almost 180 degrees due to the requirements of its intended application. Using a more directional speaker would improve the quality of the results. A picture of the Philips AD0163 is shown below.
Figure 2: Philips AD0163 Speaker
Like the speaker, the receiver must also be directional, so as not to receive echoes from unintended surfaces. It should be sensitive enough that the echo of the emitted pulse can be received after reflecting off a surface at least 40m away.
The microphone used was a basic Electret microphone with a flat frequency response to at least 15KHz, it also has an internal FET pre-amp. To make the microphone more directional, it was placed in the top end of a plastic water bottle that had been cut in half. Foam was placed around the cone created by the water bottle in order to dampen the signal from the speaker. A picture of the microphone is shown below.
Figure 3: Electret Microphone
In order to distinguish the received echo from the ambient noise, the signal must be filtered using a bandpass filter centered at 10 kHz. This filter must have a bandwidth equal to twice the inverse of the pulse duration, as determined by Fourier analysis. Since the pulse duration is 0.5ms, the necessary bandwidth is 4 kHz. The filter must also incorporate a large gain in order to amplify the signal enough that the echo can be identified.
The filter was created using four operational amplifiers. The four stages of the filter are as follows
The schematic for the implemented circuit is shown in Appendix A, along with its simulated frequency response. NE5534AP op amps were used in the filter stages because they are designed for audio frequency pre-amplifier applications. They have a wide power bandwidth, low noise, they will drive a low impedance, and they are unity gain stable. The gain stages were implemented with available op amps, but any op amp with a high enough gain bandwidth product could be substituted in their place.
Butterworth filters were chosen due to their maximally flat magnitude response in the passband with minimal ringing, and their fast roll off. However, almost any type of filter would have worked for this experiment.
The Butterworth filters were designed using a DOS software program called Active Filter Design Plusı, version 3. Unity gain was used in the filtering stages because for unknown reasons, when the circuit was designed with the gain and filtering stages combined, the circuit was unstable. This instability contributed to cutoff frequencies which were off by a factor of 10. Interestingly enough, this circuit with combined gain and filter stages performed correctly when simulated with two different simulation programs.