The mechanical design started out by investigating the possibility of 3d printing the housing. This ultimately became infeasible and a sheet metal design was used instead. From a thermal perspective, the 3d printed housing was a worse case scenario so in moving to the all steel housing, the model wasn't updated. The material used in the housing was chosen from the results of a material test for corrosion.
For the electrical design, various chips were tested but the final method for creating the signals was a square wave generator and pulse shaping filters to convert the digital signal into analog. Various power chips were tested to produce the required voltages for the design with minimal noise injection. The original modulation scheme was to be PSK but was switched to FSK for practicality for demodulation practices.
The software design focused on the framing, the protocol to accommodate multiple nodes, investigating multiple microcontrollers, and looking into requirements for compression and error correction in order to meet the data rate requirements.
The housing build was delayed by several months in redesigning in a lean fashion. A parabolic dish was designed and 3d printed in order to increase the gain of the system. Due to the power amplifier overheating consistently during tests, a short term heat sink was manufactured and another thermal model developed. The housing was tested in 15 feet of water and had minor leakage issues. The reason for this was that the hole pattern on the housing was adjusted when the seams were welded. The parabolic dish was never tested due to issues with the speaker but should be functional.
The PCB build was completed and tested at a subsystem level before integration was attempted. Each power conversion circuit had its output tested for noise and ripple voltage level. The filtering on the output of the converters was adjusted to provide as little noise as possible. The signal generation system was then tested to make sure that the signal made it from the square wave generator, through the pulse shaping filters, and then to into and out of the power amplifier circuit without much noise or distortion being added to the signal. In the end there was very little difference between the signals generated by the PCB and a signal generated by a signal generator in the lab. Finally the speaker was tested at the designed frequencies. It was determined that the speaker's frequency response was not adequate for the frequencies deemed necessary to hit our data rate requirement. The speaker was then tested at lower frequencies in order attempt a distance test, but the speaker would often not produce a usable signal, and this test was never performed.
Currently the system is set up to communicate in a unidirectional manner. One system can send a signal while the other system can receive and decode the sent signal. This proved that the send and receive modules are functioning correctly. The error correction and detection package was implemented and functioning properly along with the 2:1 compression package. Two way communication is implemented, but is however, not tested. In order to test this, another ADC will need to be purchased as one is not functional at the moment. On the modulation side, another MUX will need to be purchased in order to accommodate the ground signal needed to recalibrate the signal for the fft.
To-do list in summary:
-The speakers need replacing.
-One ADC needs replacing.
-The front plates need to be redone to the hole locations on the housing.
-Faster FFT needs to be implemented.
-Better FFT synchronization.
-Larger mux (we recommend 8:1).
-2 way communication control module.
-Power switch needs to be implemented.
-ADC switching needs to be implemented.
-Battery indicator circuit needs to be integrated.