Gregory's Server@AkaSci no clue about the problem's root cause, but it is worth pointing out that the convolutional encoder turns a constant sequence of 0s or 1s into an alternating sequence 010101. With the way this has been communicated in public media, it is not clear if the project is receiving 0101 as convolutionally encoded symbols or at the Viterbi decoder output.
@destevez
@NSFVoyager2
That does make more sense that if the serial interface line is stuck at 0 or 1, then the Viterbi encoder is generating 0101 symbols.
Otherwise it is difficult to propose a hypothesis that would cause baseband bits to follow the 0101 pattern.
The other crazy hypothesis I had was that the data line is for some reason floating and picking up the signal from the clock line, while the conv. encoder samples the data on both edges of the clock.