Thanks, but I think that approach is a bit too low level for my liking. The cable has worked fine for other applications so I'll leave it. Either I will build a companion board for the decoder output with a maxim max232 to make the levels fully compliant OR I will test with some other usb-serial adapters. I have one Aten UC232A in my hands now that I will test later.

EDIT: The Aten cable did not work either...
EDIT2:
Well, after a bit more of thinking I cannot come to any other conclusion that the serial comm is actually flawed in this design. It does not comply to RS-232 because of wrong levels, and it also does not comply to serial TTL communications because the signal is inverted. So using a proper USB-serial TTL cable is not going to work either. At the present state the design is up to luck if you happen to get a standard USB-serial adapter that is able to read 5 V levels where they ought to be +/- 12V. I haven't found one. If you do then all is well...
I'd like Howard to confirm or deny my findings if possible.

I have two options. Either I get a proper USB-serial TTL 5V adapter (like this one) and patch it in on the TX line before the inverter. It does not come cheap, but as a bonus I can use the +5V from USB line to power the whole decoder. All with one cable.
The other option is to use a maxim max232 chip to do conversion to standard RS-232 levels. Need to get the signal before the inverter here too. Like this.

It would be nice if someone could give some feedback to this.