Keeping a substantial (~22uF or so) capacitance at the Vcc and Gnd leads of the MAX232, with its 1 uF cap's, will require that the cap charge and dischargethrough the resistors. Since most folks use the MAX232A or later, which use smaller cap's, I suspect the MAX232 doesn't appear often in new designs. The larger cap should deal with the periodic demands of the MAX232 without impacting the global supply. It smooths out the ripple caused by the charge pump, and if it's a bit "starved," well, it will still be adequate. RS232 doesn't need much in excess of +/- 5 Volts in most applications.

However, I like the notion of keeping the MAX232 close to the on-board regulator, if there is one, as it's probably fast enough, if it's not an LDO, to compensate for some of the MAX232-generated noise. The requirements of the serial line, using RS232, is pretty minimal, however, so I'm not going to worry about the demands of the MAX232 relative to the rest of the circuitry.

It's already been suggested that a ferrite bead might be better than just a resistor. Resistors are adequate, however, and most guys have a few of them lying about so they can at least try it. Building a trial circuit is, of course, far from designing one for production.

RE