So I borrowed a PL2303X-based USB-COM cable (with a for me unknown manufacturer's RS232-level convertor labelled ZT213ECA), downloaded the current driver from prolific.com.tw and tried on a board populated by (T)89C51RD2 (the Temic version, but labelled Atmel) at 22.1184MHz with ST232B as RS232 level convertor.

Problems confirmed.

Autobaud at 115200Baud sucessful, answer to :020000050800F1 (bootloader version) unrecognized (Flip ends with "Timeout" box). Flip closed, time to run a decent terminal program (term by Bray). Echo of the command is OK, but the answer (which should contain 2 ascii digits, a dot and CRLF) contains garbage: 26 93 A9 0A (hexadecimal). Time to get the oscilloscope. Result shown below, the prolific chip does not recognize startbits if they come a few us earlier than expected (obviously it starts looking for trailing edge after a full stopbit time - this is confirmed by setting to 2 stopbits (btw. this is a completely erratic behaviour for an UART; the receiver should not be affected by the stopbit setting at all)).



I also tried P89C51RD2H in the same board, but it was even worse - even the echo of the command was corrupted.

It is strange it is manufactured at all - this cable is completely useless; the only situation I can think of where it will work reliably is, if the UART transmitter uses 2 stopbits while the USB virtual-com is set to 1 stopbit.

Jan Waclawek