Underwater sensor networks have a wide range of applications, many of which do not tolerate long delays and time-consuming retransmissions. Moreover, as most sensors are battery powered, saving energy is fundamental to increase network lifetime. In this paper, we investigate the minimum energy consumption for transmitting one bit of information over an underwater acoustic channel. Considering the use of binary FSK modulation and convolutional error correcting codes, we optimize the operation frequency, SNR and code rate when a target FER must be achieved. Our results show that proper choice of code rate has a great impact on the overall energy consumption and hence the lifetime of the underwater communication devices. In addition, the correct choice of the code can significantly increase the transmission range.