The Cab1 sends 24-bit commands out its serial port, via the SER2, to the Base3. As simple as it gets. I monitor what's going on using a LCS WiFi (on another LCS port) and I haven't (yet!) seen a command keyed into the Cab1 handheld that is not echoed correctly to the LCS WiFi Monitor.
The SER2 wraps the 24-bit commands in PDI packet frames for sending to the Base3 and a firmware error could mess up the commands enroute to the Base3, but that's highly unlikely.
The most likely problem source is the Base3's roster misconfiguration that is causing logic in the base to translate a 24-bit TMCC command to a Legacy command. If you have a LCS WiFi and a PC, recording the Base3's output of a problem session would help understand what's going wrong.
Good to know @Tracker John ! However I have removed my cab1/tmcc base entirely from the picture.
This does not explain how a LC+2.0 in TMCC mode can operate with the same settings and no control issues but a TMCC made 10+ years ago is erratic (it works, but better be prepared to go catch it when it flies off your layout). Second, this does not explain why iCAB app (the one released for LCS) is able to safely control the same ERR based TMCC units, with the base3.
This has to be something as simple as a signaling rate adjustment or bad instructions being generated. I mean if I have to have an entire roster of "Diesel" engines in TMCC mode with "No Sound" selected, despite being electric and steam, I might be okay with that but the functions are inconsistent.
There are some specific key sequences that apparently have a problem with the CAB3, they've been discussed and are already in the queue to be fixed.
So looks like a CAB1L handheld until that is hammered out. Thank you @gunrunnerjohn