I'm not surprised that you don't like the sound after bouncing a 24/48 file down to 16/44.1 in one step -- when you do it in one step, your dither is not working properly, since you're applying dither before the SRC takes place, which negates the effect of the dither. So, your final mix ends up as a truncated 16 bit file!
What you should do (if you're doing it all in PT) is to take your 24/48 mix, and bounce to disk to a 24/44.1 file. Then, open a 24/44.1 session, import that bounce, and either bounce it again with dither to a 16-bit file, or use the export regions command (which uses built-in dither and is quicker).
That will make a HUGE difference in your sound quality -- truncated vs. properly dithered is a pretty night and day difference. I haven't compared SRCs myself, but I would bet that doing the PT SRC correctly would put it on a much more level playing field with some of the other options out there.