My guess is if you check the signal strength of the DCS signal out of that TIU channel, you're likely to find it's lost a little of it's "punch". The Rev. L has a habit of slowly losing DCS signal strength on a channel as gates in the 74ACT244 DCS driver chip die due to transients.
Do you have the amplitude adjustment mod on the TMCC Buffer? That aids in minimizing the DCS interference from the TMCC signal.