Yes, Allison's solution would certainly work. The single diode in series with R52 does cure the spurious TX output at the beginning of transmission. For discharging the TX line, just one of the 2N3904s and one resistor (half of Allison's circuit) would do the trick for the TX to RX transition.
As a followup to the keying shape issue, I found that changing C1 to 2.2 uF gives quite nice keying shape with about 3 ms attack and 6 ms decay, much better than the original 0.5 ms attack and decay that made it a key click machine!
73,
Bob N1KW
