Hi all,
Thank you for your indulgence. I added a single transistor buffer to the FTDI Tx line, and I used the FTDI utility to invert the sense of the Tx line. Unfortunately it didn't work. This is basically Q1 R10 and R11 from Alexander Meier's circuit.
In desperation I ordered a USB interface from 409shop. I am sure it will arrive in a few days. And in fact, ordering this interface has taught me something after all. :)
In the meantime, today, I added a second layer of transistor buffer to the first, and turned the Tx inversion off again. This would be Q3 and R13. I also removed the diode between Tx and Rx. I am pleased to say that it works! I can now read the VX2r memories using VX2 Commander under Wine in Linux. I have a CSV file of my radio's contents. Chirp also does the Right Thing, and gives me a nice progress bar before it barfs with a checksum error.
I am wondering if I go back to the single-transistor buffer but with no base resistor and a 47k pullup instead of 10k would it work? I may try that, especially since I can't box up what I have as it's a spaghetti mess now, and it would be nice to just have to add one transistor and one resistor. I have not added any buffering to the FTDI chip input.
In my intermediate testing I have been trying various combinations with and without the diode, to see if it could be eliminated. My next iteration would be to drop the supply voltage to 3V3 as I'm still worried about that. I will use the 5V interface for my FT817. Incidentally, the 3V3 version worked with the 3V3 serial port on a D-Link ADSL modem which I plan to repurpose as an APRS gateway, but it has separate Tx and Rx lines, not a shared one like the VX2r.
Anyway, after this, it's just software...
73,
Andrew ZL3AME