It could have been an FCC problem. NTSC sound is at 4.5MHz (whereas the rest of the game was probably below 1MHz), and this might have required a specially shielded area on the board.
I remember talking to Jim Heller at Atari back in '79 when I was inquiring about getting composite video from the 2600. He told me it was a last minute decision in the design to have the sound come from the tv, as the engineers felt it would be more realistic.