I am reminded of a small snippet :

A professor is doing research on what stuffs inebriate.

He mixed soda with whisky and got drunk.

Next day he tried soda with rum and got drunk.

And concluded the next day that soda is inebriating with whatever it is mixed with.... :-)

4.4V and clean communication is one such. Unless the other "higher voltage" that you refer to is well above 7.5V or so !

Most of the garbage in RS232 comm is borne out of a mismatch in the timings at Tx and Rx ends. Check by altering Timer reload values. A marginal change is all that is required. And maybe your crystal is not doing its job OK.

Raghu