Yesterday I found a minor bug in the firmware. When the translator saw the NEC power signal, then saw some other signal before the receiver stopped seeing anything, an odd state was reached. The translator would have its LED stuck on, and it would never transmit the RC5 power signal. I looked into it today, and found the cause. Every time a code was received, we checked that the signal was still correct. If it wasn't, the LED wasn't turned off and nothing was transmitted. Therefore, the last code in a burst had to be correct for translation to be completed. That's not the best behavior though, since once we've seen a power code we know the button has been pressed, so we should transmit as soon as we can. The fix was very simple, and has been pushed to the repository.
In other news, the parts to build a more permanent unit should arrive on Monday. I'll be able to get this thing off of a breadboard, onto some perfboard, and into an enclosure!