The difference is an inversion of one of the hall sensor signals. Pretty simple to recognize and compensate for in software. Seems to me that both the Xie Chang and the Sabvoton that I've used detect it and work either way. I've never messed with it on any of the motors I've used. It is an interesting question.