After half a year of thorough testing the locomotive seems to be working well. Nevertheless, there are some issues. I was at first worried that the voltage drop over the two transistors would make the locomotive significantly slower than the stock one. It is slightly slower but not by much. The actual battery voltage influences the maximal speed much more. I've left the original MCU inside to play the engine sounds and both the MCUs share the input signal from the button. But due to different button debouncing implementations it sometimes happens that the locomotive is moving without making any sound and vice versa. The engine drivers got accustomed to it and now pay no attention to a stopped locomotive making sounds as if it were actually driving. In future versions this should be addressed, e.g. by removing the original MCU and handling the sound in the additional one.

I have some new ideas and plenty of old ones that might be implemented in next revisions of this project...