Conclusion

The amplifier design revealed to be a major challenge, being an excellent source of knowledge for acquiring an extended background on analog and digital electronics. Considering the multiple variables that may influence the performance of the analog audio-amplifier, all major goals were accomplished. An integrated valve amplifier with excellent characteristics has resulted. In part credit should be given to available electric simulators that allowed a much easier optimization procedure.

Due mainly to parasitic oscillations and noise problems that occurred during implementation, some of the work was delayed in a way that made the schedule to be overextended. Almost all the problems found, although undetectable to the hear, were solved. However, the digital part was delineated and implemented with success, bringing a very comfortable mean of adjustment to an excellent sounding amplifier, while the user is listening and appreciating a good sound experience. The user does not need to worry about biasing the valves as they age with use. Moreover, it may serve as an alert system for possible destructive damages, preventing major harms to the amplifier by warning the user for the need of substituting the defective devices.

Manual calibration of harmonic distortion, with the aid of a measured THD reported by the microprocessor, is working properly. Nonetheless, since the measurement is made at the primary and not at the transformer secondary, this value is not matching the one measured by the Network Signal Analyzer. The transformer itself imposes a good degree of distortion, but, curiously, for the present design, nonlinearity is somehow balanced by the amplifier. The next obvious step would then be to measure THD at the output; unfortunately time limitations did not permit this experiment.

Future Directions

While working the dissertation, some gaps were detected in the area of valve amplifiers that could be further investigated. Judging by the experience obtained from this work, the existent triode valve SPICE models have a good degree of accuracy, but pentode models can probably be improved. Still, the biggest deficit in simulation models, in the context of valves circuits, is the output transformer. In fact, after quite a searching, simulation with the use of output models is very unusual within the valve amplifier designer realm. Very few models are available and probably, if research was made around this subject, good models for these devices would become popular both to transformer and amplifier designers.