If I were to guess, given the tolerances in timing for USB 2.0, one line was slightly longer than the other on the board. That'd mean that the transition on D-

If I were to guess, given the tolerances in timing for USB 2.0, one line was slightly longer than the other on the board. That'd mean that the transition on D- would happen at a different time than the one on D+ causing corrupted data.