Gia,

You did not specify it, but I assume you are using a traditional 12-clock '51 derivative.

There is a cca. 1% mismatch between your transmitter and receiver bit-timing routines, due to different cycle counts of MOV C,bit and MOV bit,C . For example, if you use 11.0592MHz crystal, your transmitter transmits at 9600 baud, but your receiver expects 9701 bauds.

This, together with the fact that MOST of the logic-less four-leg optocouplers ARE too slow for anything but moderate baudrates^*, puts your design on the edge. No wonder you transmit 01h correctly but 02h etc. not (slight baudrate mismatch exhibits itself on the higher bits).

A single NOP should help a bit, but I recommend you to replace the optocoupler, too (see the footnote).


JW


---
* don't believe the datasheets - or, more precisely, read carefully under which conditions were they obtained - each volt and milliampere counts! To get 19200 from a PC817 I had to do various magic and well-chosen incantations and some huge collector current to the phototransistor and suffer decreased C-E voltage swing (original Sharps are very slightly better than the copies). One day we can discuss this further, but fornow for a safety margin, I recommend you to use a "digital" optocoupler, such as the H11L1 and similar, which guarantee you some 1Mbit/s or so.