It depends somewhat on the data format that you intend to use. If you use a short word, i.e. 7 bits + 1 start and 1 stop, you can probably "get by" using the same frequency for both.

The error, if you try to hit the midpoint between the two, is ~(4.167)%. If you use an oscillator of 11.5198 MHz, rather than 11.0592, and use the same setup for both baud rates. That allows you to "set it and forget it".

There is a gotcha. The notion of "getting by" is based on the assumption that you use only 9 bits per character, then start over, as async protocol would dictate if you use 7 bits + 1 start and 1 stop bit</n>. If you do that, then you have 9*.04167 bits error per character, which is less than a half bit, so you'll lose no characters. If you use 8 bits, as in N-8-1, you'll still have less than half a bit of error for each character, which will mean you still "get under the flap." If you use an 11-bit format, you have .4583 bits of error, and that's a bit close ...

This whole thing falls apart if your remote station is not set at the exact frequency that's required for its baud rate. If it's producing exactly 9600, as it would in the typical 805x case with an oscillator of 11.5692E6, or exactly 10400, as it would using 11.5198E6, then you're fine. These computations assume an oscillator of the typical 100ppm quality. Otherwise, all bets are off.

RE