Hy
If you accept same probability for every floating input state (i'm not sure of that)
the chance is 1/2 for wrong direction of translator chip (otherwise the pull up on PTA1 works fine), then 1/2 for wrong state of translator input.
therefore the chance is only 1/4 to get a trouble (not entering monitor state)
may that explain the problem was not seen ?
bye
and thanks for your answers