Of course I'm obsessed with the possibility of a 2-wire self-powered PS2 version. To that end I reduced the 470uF cap to 220uF as you have in your size-matters module. I also added a 4.7K resistor across the 5V 78L05 output to simulate the ~1 mA load of a PIC. Then I triggered it via the 2-wire coupler with both 18V command voltage and then only 12V AC on the track. As scope photo above shows using short PS2 coupler pulse, 220uF is enough to power ~6 bursts at command voltage! And even with a lower 12V track voltage it still put out 3 bursts...and reliably activates the receiver. In other words, I think your 100 msec TX time which would be ~3 bursts is long enough.
So, again, I think by simply parallel'ing your 2 track power inputs and a trigger input pair, there is enough energy in the 220uF to release a valid burst in 2-wire self-powered mode! I'm undoubtedly the only person who finds this exciting...but I've learned to be thankful for the small things in life! 
